The Hydrogeology and Hydrochemistry of the Mt. Tom Price Mine, Pilbara, Western Australia – A Groundwater Flow Model.
______
A thesis submitted in partial fulfilment of the requirements for the degree of Master of Science in Engineering Geology with Honours in the University of Canterbury by NEIL MANEWELL ______
University of Canterbury September, 2008 ABSTRACT
The Mt. Tom Price Mine, located in the Pilbara region of Western Australia, has been the site of major iron ore mining since the 1960s by Rio Tinto Iron Ore/Pilbara Iron. The thesis project area covers approximately 121 km2, covering the Mt. Tom Price Mining area and the surrounding catchment boundary. The climate in the Pilbara region is arid, with rainfall driven by seasonal cyclonic events, producing 300 mm/year net rainfall on average. The geology of the Mt. Tom Price area consists of a series of banded iron formations (BIF) and shales that are generally low in hydraulic conductivity values. Iron ore in the region is produced through the process of supergene enrichment whereby gangue minerals are dissolved and replaced with haematite and goethite. Mining is focused in a series of open cast pits including, North Deposit, West Pits, Centre Pits, Southern Ridge, South East Prongs, Section Six, Section Seven, and the proposed Marra Mamba Pits.
Due to the impermeable nature and complex geology of the BIF sequence, groundwater flow is dominated by bedrock aquifer flow, with compartmentalization occurring in several areas of the mine. Highly faulted and folded units can also have increased hydraulic conductivity values. Pit floor lowering began to encounter the regional water table in early 1994. A series of dewatering bores and depressurization measurements have been utilized to ensure dry mining practice. This data was used to help understand regional groundwater flow and create the Mt. Tom Price Groundwater Model (MTPGM).
A 3D geological model of the project area was created to aid visualisation of semi-regional hydrogeology. From this model, accurate template files were created so that geological detail loss is kept to a minimal when entering hydrogeological parameters into the MTPGM. The MTPGM was setup using PMWIN Pro, a graphical user interface for use with MODFLOW. Stresses such as recharge and pumping were entered via software packages within MODFLOW. The model was run to simulate measured 1994-2007 responses to dewatering and high rainfall events. A Parameter Estimation (PEST) software package and trial and error calibration was used to lower stress response variances that were observed in the model output files. This was achieved by the adjustment of hydrogeological parameters such as hydraulic conductivity and specific yield values. A prediction simulation of final pit lake recovery was created Using the calibrated MTPGM. Recovery curves predicted that full recovery of the water table of the pit voids varied from 96 to 120 years, recovering to levels close to the initial heads measured in 1994 before large-scale pumping commenced.
The hydrochemistry of the groundwater in the mining area is highly influenced by geological hosts, with clearly defined hydrochemical signatures approximated for each screened geological unit. Due to the sulphur rich, acid- forming Mt. McRae Shale, regular monitoring of pit and groundwater is essential. Final pit lake water quality was estimated using final pit levels and recovery rates approximated from the MTPGM, combined with historical data and previous groundwater quality reports. Pit lake water quality is dominantly driven by evaporation concentration, caused by high evaporation rates and low throughflow. Pit waters are expected to be brine waters (>100,000 mg/L TDS), with high levels of acidity values occurring in the South East Prongs and Section Six pits due to the exposure of the acid forming Mt. McRae Shale above the pit lakes at these localities.
Future studies should focus on more detailed modelling of the compartmentalised aquifer systems. This would produce much more accurate final pit lake levels. Further study of the Mt. McRae Shale formation and its implications on acidity should also be undertaken. Seasonal fluctuations in lake levels will affect acidity due to the continual re-exposure and oxidation of the Mt. McRae Shale. This could be studied to help understand short term pit lake quality conditions and help to predict long term acidity conditions in the pit lakes.
i
ACKNOWLEDGEMENTS
Firstly, I would like to thank Rio Tinto Iron Ore and Pilbara Iron for their support throughout this thesis. Big thanks go out to George Domahidy and Scott Rathbone, who helped organise this joint venture with University of Canterbury. They also provided continued help during the project duration, and directed me to relevant help and resources. Wade Dodson was also helpful in reviewing groundwater modelling ideas and providing tuition.
I would like to thank David H. Bell for his work also setting up this project, as well as continued support and communication throughout the year. Travis Horton has also been helpful towards the end of this project.
Huge thanks go out to Catherine Moore of Lincoln Ventures Ltd. for her continued help with groundwater modelling, providing tuition and direction. She devoted many hours of her busy schedule on this often challenging groundwater model.
I would like to thank Kathryn Rozlapa of Aquaterra Perth for her tuition and help with setting up the groundwater model. She also provided helpful electronic and phone communication throughout the project duration.
Thanks go out to the onsite Tom Price Hydrogeologists for their devoted help during fieldwork, as well as providing data via electronic communication. Cheers: Emma Gallagher, Lee Evans, Tim Kendrick, Lindsay Campbell, and Chris New.
The Tom Price Geotechnical Team also provided much welcomed support. Big thanks to Cameron Boyle, Manni Mehu, and the Tom Price onsite Technicians.
I would like to thank the other geology postgraduates for their support throughout this 18 month project. Cheers: Henry, Richard, Kirsty, Myfanwy, Greer and Jeremy for providing the entertainment.
Lastly I would like to thank my family, especially my Mum, Dad and sisters for their support and guidance throughout the duration of this thesis.
ii ABBREVIATIONS
General EDMS Environmental Database High P High Phosphorus KH Horizontal hydraulic conductivity KV Vertical hydraulic conductivity Low P Low Phosphorus MGA94 Map Grid of Australia 1994 mRL Mine Reference Level (m) MTPGM Mt. Tom Price Groundwater Model MTPPGM Mt. Tom Price Prediction Groundwater Model PEST Parameter Estimation Piezo Piezometer observation hole PMWIN Pro Processing Modflow Professional S Storage Coefficient S/Pipe Open standpipe observation hole SY Specific Yield TDS Total Dissolved Solids TPMG Tom Price Mining Grid
Geological Units BIF Banded Iron Formation DET Detritus DG Dales Gorge Member FWZ Footwall Zone JER Jeerinah Formation JOF Joffre Member MCS Mt. McRae Shale MTS Mt. Sylvia Formation MM Marra Mamba Iron Formation WBS Whaleback Shale Member WT Wittenoom Formation
Pits CTR Centre Pit MME Marra Mamba East MMW Marra Mamba West NEPX North East Prongs Extension NTD North Deposit SEP South East Prong SSIX Section Six SSEV Section Seven STR Southern Ridge WEST West Pit
x
TABLE OF CONTENTS
CHAPTER 1 INTRODUCTION...... 1 1.1 Project Background ...... 1 1.2 Location & Extent ...... 1 1.3 Rainfall and Climate ...... 2 1.4 Mining ...... 3 1.5 Structural Setting ...... 3 1.6 Project Objectives ...... 8 1.7 Thesis Format ...... 8
CHAPTER 2 GEOLOGY & HYDROGEOLOGY OF THE MT. TOM PRICE MINE AREA ...... 10 2.1 Introduction ...... 10 2.2 Fortescue Group Formations ...... 12 2.2.1 Jeerinah Formation ...... 12 2.3 Hamersley Group Formations ...... 12 2.3.1 Marra Mamba Iron Formation...... 12 2.3.2 Wittenoom Formation ...... 12 2.3.3 Mt. Sylvia Formation...... 13 2.3.4 Mt. McRae Shale...... 14 2.4 Brockman Iron Formation ...... 14 2.4.1 Dales Gorge Member ...... 14 2.4.2 Whaleback Shale ...... 15 2.4.3 Joffre Member ...... 15 2.4.4 Yandicoogina Shale Member ...... 15 2.5 Mt. Tom Price Ore bodies ...... 16 2.6 Banded Iron Formation (BIF) Genesis and Characteristics ...... 16 2.6.1 Supergene Enrichment...... 16 2.7 Hydrogeology of Mt. Tom Price...... 18 2.7.1 Groundwater Flow...... 18 2.7.2 Dewatering History ...... 22 2.7.3 Piezometer & Monitoring Bore Network...... 23 2.7.4 Dewatering Target Areas...... 24 2.7.5 Groundwater Management ...... 26 iii
2.7.6 Slope Depressurization...... 27 2.7.7 Surface Water Management ...... 27 2.8 Synthesis...... 28
CHAPTER 3 GEOLOGICAL MODEL & CONCEPTUAL GROUNDWATER MODEL ...... 29 3.1 Introduction ...... 29 3.2 Rio Tinto Drillhole Database...... 29 3.3 Historical Geological Cross-sections...... 30 3.4 Vulcan Triangulation Surfaces ...... 33 3.5 Construction of Geological Model ...... 33 3.6 Conceptual Groundwater Model ...... 35 3.7 Relationship between Conceptual and Numerical Models...... 39 3.8 Synthesis...... 42
CHAPTER 4 MT. TOM PRICE GROUNDWATER MODEL ...... 43 4.1 Introduction ...... 43 4.2 Previous Groundwater Models ...... 43 4.2.1 South East Prongs Dewatering Model...... 44 4.2.2 North Deposit Dewatering Model ...... 45 4.2.3 Section Six Storm Water Storage Model...... 46 4.2.4 Pit Void Closure Modelling...... 48 4.3 MTPGM Model Setup ...... 48 4.3.1 Model Input Data...... 49 4.3.1.1 Initial Model Setup...... 49 4.3.1.2 Boundary Conditions...... 49 4.3.1.3 Initial & Prescribed Hydraulic Head ...... 51 4.3.1.4 Grid Design ...... 53 4.3.1.5 Time Discretisation ...... 55 4.3.1.6 Model Parameters...... 55 4.4 Modflow Packages...... 57 4.4.1 Block Centred Flow (BCF) Package ...... 57 4.4.2 Flow Packages...... 57 4.4.2.1 Recharge...... 57 4.4.2.2 Wetting Capability...... 58 iv
4.4.3Well Package...... 59 4.4.4 Solvers ...... 59 4.4.4.1 Preconditioned Conjugate Gradient 2 (PCG2) Package...... 60 4.4.5 Output Control...... 60 4.4.6 Head Observations...... 61 4.5 Model Calibration ...... 61 4.5.1 Introduction ...... 61 4.5.2 Results ...... 62 4.5.3 Calibration Considerations ...... 63 4.5.4 Trial and Error Calibration ...... 64 4.5.5 Parameter Estimation (PEST)...... 65 4.5.6 Final Calibration...... 67 4.6 Sensitivity Analysis &Model Prediction Uncertainty ...... 70 4.7 Predictions...... 71 4.7.1 Prediction Model Setup ...... 71 4.7.2 Results ...... 72 4.7.3 Pit Infilling...... 78 4.8 Model Limitations ...... 79 4.9 Synthesis...... 80
CHAPTER 5 MT. TOM PRICE HYDROCHEMISTRY ...... 81 5.1 Introduction ...... 81 5.2 Previous Hydrochemical Work...... 82 5.2.1 Aquaterra Hydrochemical Modelling...... 82 5.2.2 EWL Sciences Water Quality Assessment...... 83 5.3 Mt. Tom Price Hydrochemistry...... 84 5.3.1 Background...... 84 5.3.2 Time Series Water Quality Data...... 85 5.3.3 Acidity Changes Over Time...... 87 5.3.3 Discussion...... 88 5.4 Chemical Data Presentation...... 89 5.4.1 Piper Plots...... 89 5.4.2 Stiff Plots...... 91 5.4.3 Discussion...... 95 v
5.5 Future Predictions...... 95 5.6 Synthesis...... 97
CHAPTER 6 SUMMARY AND CONCLUSIONS ...... 98 6.1 Projects Aims & Objectives...... 98 6.2 Mt. Tom Price Hydrogeology...... 98 6.3 Mt. Tom Price Geological & Conceptual Groundwater Models ...... 99 6.4 Mt. Tom Price Groundwater Model...... 99 6.5 Pit Void Closure Implications ...... 100 6.6 Hydrochemistry Conditions & Quality Prediction ...... 101 6.7 Future Investigations ...... 103
REFERENCES...... 104
APPENDICES...... 106
vi
LIST OF FIGURES
Figure 1.1: Location map for the project area illustrating the catchment bounds of the project area and current mining areas ...... 2 Figure 1.2: Tom Price daily rainfall sampled within project area from 1998 – 2008 ...... 3 Figure 1.3: Simple stratigraphic column of the geological groups and formations in the region ...... 4 Figure 1.4: Geological map displaying regional structure of the Mt. Tom Price area. Project area outlined...... 5 Figure 1.5: Cross sections CSA and CSB through the Mt. Tom Price Mine displaying general structure...... 6 Figure 1.6: Geological Map of the Mount Tom Price mine area illustrating mineralisation zones ...... 7 Figure 2.1: Stratigraphy of the Hamersley Group including details of the Marra Mamba and Brockman Iron Formations...... 11 Figure 2.2: Geological map of the Mt. Tom Price mine area displaying geological structure and mining locations...... 17 Figure 2.3: Simplified model of supergene enrichment ...... 19 Figure 2.4: a) Feb-May 1994 water levels and Tom Price geology. b) June 2007 water levels and Tom Price geology ...... 21 Figure 2.5: Mt. Tom Price water bore and current (June 2007) piezometer locations ...... 25 Figure 2.6: Section of horizontal drain holes in pit wall ...... 27 Figure 3.1: Aerial view displaying complete Mt Tom Price drillhole database...... 30 Figure 3.2: Geological Map of the Tom Price mine area displaying sampled drillholes (red) and their corresponding cross-sections ...... 31 Figure 3.3: Example of interpreted cross-section (looking west through CS1) drillhole values entered into AUTOCAD...... 32 Figure 3.4: Map of cross section localities from 1972 1:4800 survey. Coordinates are displayed in the old Tom Price mine grid format ...... 32 Figure 3.5: Cross section example through 1800 E (15300 E) (looking east ...... 32 Figure 3.6: Oblique 3D view displaying a sample of Vulcan triangulation surfaces ...... 34 Figure 3.7: Geological Model (map view)...... 34 Figure 3.8: Oblique 3D view of geological model ...... 36 Figure 3.9: Conceptual groundwater model through main areas of hydrogeological interest 38 Fig 3.10: A) Conceptual cross-section through 15720 E (looking east), displaying different hydrogeological units. B) Cross-section through the Mt Tom Price groundwater model at 15720 E ...... 40 vii
Figure 3.11: Oblique 3D view in the modelling environment (looking north-west) displaying Interpreted geology ...... 41 Figure 3.12: Sample of geological template file between 720-670 mRL in map view (Layer 7-8)...... 41 Figure 4.1: Final head values at last stress period (2014) with 1999 pumping abstraction rates. Black line represents final pit floor levels...... 45 Figure 4.2: Areas of predicted groundwater levels below 630 mRL (Oct 2004) ...... 45 Figure 4.3: North Deposit dewatering requirements displaying simulated waterlevels vs. pit floor progression ...... 46 Figure 4.4: Section Six Particle travel over 1000 years following discharge of excess water from intense 1:20 year flooding (160 ML) ...... 47 Figure 4.5: Mt. Tom Price model coordinate system values for use in PMWIN Pro...... 49 Figure 4.6: IBOUND array of Mt. Tom Price model in PMWIN Pro. Insert displays cell values and their corresponding colours...... 51 Figure 4.7: Oblique 3D view of the Mt. Tom Price Groundwater Model with an aerial photograph of the mine superimposed...... 52 Figure 4.8: Grid Spacing required for stable simulation using MODLFOW simulations...... 53 Figure 4.9: DXF image displaying final pit designs (pink) vs. rendered initial water table levels (blue)...... 54 Figure 4.10: Hydrograph example of a moderately calibrated observation hole...... 63 Figure 4.11: Hydrograph example of a poorly calibrated observation hole, displaying large errors (up to 30 m) with a poor response to stresses...... 63 Figure 4.12: Super Parameter Selection ...... 66 Figure 4.13: Hydrograph example of a well calibrated observation hole...... 69 Figure 4.14: Hydrograph example of a moderately calibrated observation hole...... 69 Figure 4.15: Hydrograph example of a poorly calibrated observation hole ...... 70 Figure 4.16: MTPGM uncertainty displaying measured heads against modelled heads...... 71 Figure 4.17: North Deposit Pit recovery from prediction model...... 75 Figure 4.18: South East Prongs Pit recovery from prediction model ...... 75 Figure 4.19: Marra Mamba Pit recovery from prediction model ...... 75 Figure 4.20: Section Six Pit recovery from prediction model ...... 76 Figure 4.21: West Pits recovery from prediction model...... 76 Figure 4.22: Aerial view of full pit lake recovery conditions at Mt. Tom Price Mine, Circa 2130...... 78 Figure 4.23: North Deposit recovery curve from prediction model with pit infill ...... 79 Figure 5.1: MM01 hydrochemistry values vs. time...... 86 viii
Figure 5.2: Section 6 hydrochemistry values vs. time...... 87 Figure 5.3: WEP05 hydrochemistry values vs. time ...... 87 Figure 5.4: Acidity sampled at MM01 vs. time with polynomial trend line added...... 88 Figure 5.5: Piper Plot displaying hydrochemical variation according to sampled geological host – mid 2007...... 90 Figure 5.6: Piper Plot displaying hydrochemical variation according to sampled location – mid 2007 ...... 91 Figure 5.7: Mt. Tom Price stiff plots (February-August 2007 EDMS data) ...... 93 Figure 5.8: Mt. Tom Price stiff plot locations (January – June 2007 EDMS data) ...... 94
LIST OF TABLES
Table 2.1: Historic dewatering targets (mRL) ...... 26 Table 2.2: Historic Dewatering volumes ...... 26 Table 4.1: Parameter values for the SEP dewatering model...... 44 Table 4.2: Model parameters for the SSIX storage model...... 47 Table 4.3: Predicted long term pit lake conditions at Mt. Tom Price...... 48 Table 4.4: Model parameters used in Mt. Tom Price Groundwater Model...... 56 Table 4.5: Predicted long term pit lake conditions at Mt. Tom Price from June 2007 recovery scenario ...... 73 Table 4.6: Predicted long term pit lake conditions at Mt. Tom Price from Final Pit recovery scenario ...... 73 Table 4.7: Head difference correction values from transient calibration hydrographs...... 77 Table 4.8: Predicted long term pit lake conditions at Mt. Tom Price from Final Pit recovery scenario (Corrected from transient results and modelled errors)...... 77 Table 5.1: Long term hydrochemical predictions post mining ...... 82 Table 5.2: Final pit lake conditions according to pit lake levels for the STR Mine ...... 83 Table 5.3: Composition of 1:2 extract of the Mt. McRae Black Shale...... 84 Table 5.4: Final pit lake dimensions and conditions...... 96 Table 6.1: Predicted long term pit lake conditions at Mt. Tom Price from Pit recovery...... 101 Table 6.2: Final pit lake dimensions and conditions...... 102 Table 6.3: Final pit lake conditions according to pit lake levels where Mt. McRae Shale is present...... 102
ix
Chapter 1: Introduction
CHAPTER 1
INTRODUCTION
1.1 PROJECT BACKGROUND The Mt. Tom Price Mine, located in the Pilbara region of Western Australia, is the site of intense iron ore mining, of which millions of tonnes of BIF derived haematite and goethite ore have been mined since mining commenced in the 1960s. The hydrogeology and the hydrochemistry of the Mt. Tom Price mine area are not well understood. Due to the intensely folded and faulted nature of the area, groundwater flow patterns are compartmentalised within the mine footprint. As several mining areas are nearing closure, it is important that groundwater flow patterns are correctly understood in order to predict long term pit void lake levels and behaviour. The extraction of iron ore has exposed potentially acid forming units within these pit voids. The combination of the impermeable nature of surrounding rocks and final water level values will be influential upon pit lake hydrochemistry.
The main goal of this thesis is to produce a 3D groundwater flow model on which future decision making can be based in regard to environmental management decisions on pit backfilling and void closure options. This thesis also serves as a template to produce a reliable groundwater model with limited documented geological and hydrogeological data in other mining areas.
1.2 LOCATION & EXTENT Mt. Tom Price is located within the Hamersley Basin in the Pilbara region of Western Australia. This large basin is approximately 1500 km north of Perth and contains Australia’s largest iron ore reserves. These banded iron formation (BIF) derived deposits have been the focus of exploration, mining and research since the 1960s, which has since intensified due to the rapid growth of Asia’s booming economy.
1 Chapter 1: Introduction
The current Mt. Tom Price Mine covers an area of approximately 60 km2 over steep terrain with highly variable geological units and structures. The project area was chosen to incorporate the local catchment boundary (Figure 1.1). This area was selected upon analysis of topographical information and plotting boundary extents from ridge lines surrounding the mine. As a result, an area of 11 x 11 km, or 121 km2 was defined, so that an accurate calculation of precipitation input can be calculated.
WESTERN AUSTRALIA
Mt. Tom Price Mine Area
Figure 1.1: Location map for the project area illustrating the catchment bounds of the project area (red) and current mining areas (black).
1.3 RAINFALL AND CLIMATE The Pilbara climate is arid, with rainfall dominated by cyclonic rainfall events that occur in the summer months (January-March). During this period, isolated and intense rainfall events can produce precipitation in excess of 100 mm/day (Figure 1.2). Mean annual precipitation however is approximately 300 mm year (Rio Tinto Hydrogeological Database).
Evaporation rates in the region are very high, especially in the summer months during which evaporation rates can reach approximately 3000 mm/month. The largest deficit between rainfall and evaporation rates occurs during the months of July to November (Rathbone, 2005).
2 Chapter 1: Introduction
Figure 1.2: Tom Price daily rainfall sampled within project area from 1998 – 2008 (Rio Tinto Hydrogeological Database, 2008).
1.4 MINING Mining at Mt. Tom Price commenced in the 1960s, comprising of a surface drill and blast program and the use of large-scale loaders transporting iron ore directly to the onsite plant. Here, iron ore is processed and loaded by conveyor belt onto train carriages. Thousands of tonnes of iron ore are transported to the Dampier processing plant every day.
Mining at Mt. Tom Price was originally carried out by Hamersley Iron Pty Ltd. In 2004, Robe River Iron and Hamersley Iron merged to form Pilbara Iron, which has since been taken over by Rio Tinto Iron Ore (www.riotintoironore.com).
Up until 1994, mining was specifically above water table and as a result all mining machinery is equipped for dry mining of iron ore only. Since this period, a series of dewatering bores have been operational to ensure dry mining throughout all mining localities.
1.5 STRUCTURAL SETTING The rocks of the Hamersley Basin consist of the Mount Bruce Supergroup, which comprise of supracrustal rocks of Archaean to Palaeoproterozoic age, resting on a basement of older
3 Chapter 1: Introduction
Archaean granites and greenstones which form the Pilbara Craton. This craton forms the platform of the Mt. Bruce Supergroup, consisting of the Fortescue, Hamersley and Turee Creek Groups. The basalt-dominated Fortescue Group, at the base, is conformably overlain by the Hamersley Group, which comprises of banded iron formation (BIF), carbonate, fine grained siliclastics and acid volcanic rock. This group is conformably overlain by the Turee Creek Group, a sequence of siliclastic rocks that typically coarsens up the stratigraphic section (Figure 1.3) (Tyler & Thorne, 1990).
The iron ore rich deposits mined in the Pilbara derive from the Hamersley group, a 2.5 km thick late Achaean/early Proterozoic age stratigraphic assemblage. This group has been intensely folded and faulted during three main phases of shortening events (Figure 1.3). An early extensional phase took place approximately 2700 Ma, incorporating the Fortescue Group. This was followed by a post rift phase that incorporated the Fortescue and Hamersley Groups (Etheridge et al, 1996). The last event commenced approximately 1700 Ma, forming dome structures exposed at the surface and a suite of linear folds. Faulting mainly affected the south-western and western margin in the form of dextral strike slip and normal faulting (Tyler & Thorne, 1990).
2500 Ma Figure 1.3: Simple stratigraphic column of the geological groups and formations in the region. 4 Chapter 1: Introduction Mt. Tom Price Mine Area Figure 1.4: Geological map displaying regional structure of the Mt. Tom Price area. Project area outlined. Refer to Figure 1.3 key for geological formations (purple lines indicate regional faults, green lines indicate dolerite dykes) The Mt. Tom Price iron ore deposits are situated on the southern limb of the Turner Syncline, a ~35 km long structure, intercepted by numerous faults and dike swarms (Figure 1.4). As a result of further folding events, two east-west parallel ridges have formed, creating secondary synclinal structures. These synclines dip at approximately 30° with a corresponding alignment to the Turner Syncline (Gilhorne, 1975). The main, high grade ore body of Mt. Tom Price is located within the fold nose of the Turner Syncline, post dating the folding and faulting events of the upper Wyloo Group. The syncline cuts through earlier extensional faults and folds, finally being intruded in the Proterozoic by the major dolerite dyke swarm, associated with a major rifting event. These mafic dykes post-date folding as well as ore formation (Hamersley Iron, 2000). The largest fault in the mining area is known as Southern Batter Fault, which runs along strike of the northern sub-limb of the syncline (Figure 1.5-1.6), displaying normal displacements of up to approximately 100 m at some localities. This fault has generated a large fault zone south of the Centre Pit mine area, made up of BIF breccia material that extends several hundred metres deep (Figure 1.5). 5 Chapter 1: Introduction The South East Prongs Fault is located just north of the South East Prongs Pit, displaying normal displacements of up to 50-100 m. The Box Cut Fault runs along the north wall of the North East Prongs, displaying similar displacements (Figure 1.6).An unconformable contact with the Marra Mamba and Wittenoom formations in the south of the mining area implies that this contact is a large, northerly dipping fault (Figure 1.5-1.6). Little documented information is available on this faulted contact, but will likely become available with continued Marra Mamba pit studies. Cross sections have been created through the mine area to illustrate the general structure of the mining area (Figure 1.5). Detailed structural information is illustrated in Figure 1.6, where the numerous localised faults and anticlines/synclines are displayed in relation to geology. CSA CSB Figure 1.5: Cross sections CSA and CSB through the Mt. Tom Price Mine displaying general structure (Refer to Figure 1.6 for locations). Bold Italics represent mining locations. Refer to page x for definitions Refer to Figure 2.3 for eastings locations. 6 Chapter 1: Introduction N d (Refer 1.5 for CSA, CSB to Figure CSA Map uses Australian MGA94 gri CSB ical database). g eolo g dro y cross-sections (Rio (Rio Tinto H cross-sections 7 Figure1.6: Geological of areaMapTom the Mount Price mine illu strating mineralisationNB zones. Chapter 1: Introduction 1.6 PROJECT OBJECTIVES The primary goal of this thesis is to create a 3D Mt. Tom Price Groundwater flow model (MTPGM) of the mine area to help understand semi regional groundwater flow in the mining area. For this to be achieved, a complete 3D geological model should be created to help visualize hydrogeological structures and piece together geological structures that have not been studied in great detail. Using a prediction version of this model (MTPPGM), final pit void conditions including recovery rates and final pit levels elevations were approximated. These levels are indicative upon final pit lake quality post mine closure and were used to predict long term hydrochemical conditions of the pit lakes. Using the results of these datasets, combined with previous studies on groundwater conditions, final pit lake conditions can be approximated. Lake levels will be calculated by the Mt. Tom Price Prediction Groundwater Model (MTPPGM). These levels will be highly influential upon final water quality conditions as throughflow will be extremely low due to the low transmissivity of the tight BIF and shale formations. Given these objectives the following targets have been established: 1) Understand semi-regional groundwater flow in the Mt. Tom Price mining area. 2) Create a 3D geological model of the mining area. 3) Create a numerical semi regional groundwater flow model (MTPGM) using MODFLOW calculations. 4) Predict final pit lake levels and recovery rates upon mine closure. 5) Predict long term final pit lake water quality conditions. 1.7 THESIS FORMAT The organisation of this thesis is as follows: Chapter 2: Geology & Hydrogeology of the Mt. Tom Price Mine Area This chapter outlines the nature of the geological units and structure of the mining area. Hydrogeological parameters will be explored, outlining the major features that are influential upon groundwater flow patterns and groundwater quality. Documented hydrogeological 8 Chapter 1: Introduction information will be explored, such as, known flow patterns, historical and current dewatering status, and piezometer networks. Chapter 3: Geological Model & Conceptual Groundwater Model This chapter outlines the methods involved in the construction of a 3D geological model using AUTOCAD, a 3D modelling program. This will be created from data attained from historical cross-sections, drillhole data and Vulcan triangulation models. The relationship between the geological, conceptual and numerical groundwater models will also be explored. Chapter 4: Mt. Tom Price Groundwater Model Chapter 4 comprises of an introduction to groundwater modelling using Processing Modflow Professional (PMWIN Pro) and the steps undertaken to create the MTPGM and MTPPGM. Calibration techniques will be explored to produce reliable results and the errors associated with these techniques will be discussed. Using this calibrated model, final pit levels and recovery rate results will be presented. Chapter 5: Mt. Tom Price Hydrochemistry Mt. Tom Price Hydrochemistry will be explored in this chapter by analysing historical data collected from dewatering and observation bores. Data will be presented as time series plots, spatial variances, piper plots and stiff plots. Using these results, historic reports and outputs from the MTPPGM, long term pit water quality predictions will be presented. Chapter 6: Summary & Conclusions Chapter 6 summarises the major results of the project, outlines major findings and recommendations. 9 Chapter 2: Geology and Hydrogeology CHAPTER 2 GEOLOGY & HYDROGEOLOGY OF THE MT. TOM PRICE MINE AREA 2.1 INTRODUCTION The stratigraphy of the Mt. Tom Price mine area consists of a series of banded iron formations (BIF), with various interbedded shale, chert and volcaniclastic units. These deposits were formed during the Archaean-Proterozoic in oxygen deprived environments. After series of late Archaean/Proterozoic tectonic events, these formations became highly faulted and folded, with numerous doleritic dyke swarms intruding through most of the region. Selective iron mineralisation took place throughout the Brockman and Marra Mamba Iron Formations. This process is known as supergene enrichment, a process involving the augmentation of haematite and goethite content. This is caused by the dissolution of gauge material, aided by groundwater percolation and structural preparation. Hydrogeologically the region is dominated by bedrock groundwater flow through discontinuities in tight BIF and shale units. Areas of mineralisation and faulting can increase permeability values, resulting in perched water bodies and compartmentalisation of groundwater flow in the mining area. The earliest unit observed in the project area is the Jeerinah Formation, part of the Fortescue Group (Figure 2.1). This is followed by the Hamersley Group, an assemblage comprising of BIF and shale formations containing large amounts of the iron ore mined in the region. This comprises of the Marra Mamba Formation, Wittenoom Formation, Mt. Sylvia Formation, and Mt. McRae Formation. The Brockman Iron Formation is a sub-group within the Hamersley Group and has been the focus of iron ore mining over the last 40 years. This formation is the highest observable formation in the project area and is made up of the Dales Gorge Member, Whaleback Shale, Joffre Member, and Yandicoogina Shale Member. 10 Chapter 2: Geology and Hydrogeology Figure 2.1: Stratigraphy of the Hamersley Group including details of the Marra Mamba and Brockman Iron Formations (Hamersley Iron, 2000). 11 Chapter 2: Geology and Hydrogeology 2.2 FORTESCUE GROUP FORMATIONS 2.2.1 Jeerinah Formation The Jeerinah Formation is the earliest observable formation in the Mt. Tom Price mining area, forming around 2750 Ma in a sub-aerial volcanic environment (Trendall et al., 1998). This unit consists of a sequence of a number of different geological units including: dolerite, shale, dolomite, dolomitic mudstone, chert, and minor tuff. The formation measures appromatiely 1000 m thick with a conformable contact with the overlying Marra Mamba Iron Formation at most locations. At the Turner Syncline however, this contact is unconformable, likely caused by localized slipping caused by the formation of the syncline during the late Archaean (Trendall & Blockley, 1970). 2.3 HAMERSLEY GROUP FORMATIONS 2.3.1 Marra Mamba Iron Formation The Marra Mamba Iron Formation consists of unmineralised (unenriched) BIF of thicknesses of approximately 230 m. This unit is a supersequence, dated near its base at ca 2597 Ma (Trendall et al., 1998) and is the host rock for all the major Marra Mamba iron ore deposits in the region. Major iron mineralisation occurs in the upper levels of this unit. This formation can be divided into three members. The earliest is the Nammuldi Member which measures ~135 m thick where unenriched and contains chert-rich BIF and thin discrete shale bands. The overlying MacLeod Member is 35m thick, consisting of BIF, chert and carbonate with interbedded shales. The uppermost Mt. Newman Member measures 60 m thick, comprising of manganese-bearing BIF with interbedded carbonate and shale bands. The shale bands within the Marra Mamba Formation are laterally continuous and represent kaolinised volcanic tuff layers, providing excellent regional markers across the region. “Macrobands” within the formation can also be characterised by their differing radioactive log ‘signatures’ (Trendall & Blockley, 1970; Gilhome, 1975; Harmsworth et al., 1990; Blake & Barley, 1992; Krapez, 1997). 2.3.2 Wittenoom Formation The Wittenoom Formation can be divided into three stratigraphic units. The earliest, the West Angelas Member (40 m thick), contains magniferous shale, chert and thinly bedded dolomites 12 Chapter 2: Geology and Hydrogeology with minor BIF at the base. Dolomites in the formation may also display cross bedding and slumps, as well as stylolites and chert nodules (Trendall & Blockley, 1970). The middle Paraburdoo Member (up to 150 m in thickness) contains chert bands and crystalline dolomite bands. The uppermost Bee Gorge Member (~35 m thick) (~2561 Ma) contains alternating shale and dolomite bands with minor BIF, cherts and volcaniclastics (Trendall & Blockley, 1970; Harmsworth et al., 1990). The Wittenoom Formation has an inconsistent stratigraphic thickness throughout the region. Due to dolomitic solutioning, geological structure, sparse nature of the outcrop, and high weathering rates, the true thickness off the outcrop is hard to determine, but has been estimated at between 300 m and 600 m thick (Trendall et al., 1998). The distribution of the outcrop directly corresponds to the broad valleys between the hog-back ridges of the Brockman and Marra Mamba Iron Formations. This formation represents a reversion to principally clastic sedimentation within the basin, with a gradual transition to chemically precipitated sedimentation as seen in the dolomites low in the formation. These units have a limited lateral persistence, with chert and dolomite lenses measuring less than 2 m long. Due to the weak nature of the shale, the formation has a greater degree of folding as compared to the other units (Trendall & Blockley, 1970; Harmsworth et al., 1990). 2.3.3 Mt. Sylvia Formation The Mt. Sylvia Formation (30 m thick) has a conformable contact with the underlying Wittenoom Formation. This can be separated into three prominent BIF-chert members separated by shale, chert and dolomite. The formation is persistent with a uniform thickness and therefore serves as an excellent structural marker. It is also spatially restricted to the eastern part of the Hamersley Province, a rare example of terrigenous clastics in this region (Gilhome, 1975; Harmsworth et al., 1990). The lowermost unit is composed of a cross bedded siltstone (tuffaceous in places), measuring ~20 m thick. The uppermost BIF is informally known as ‘Bruno’s Band’ and can be utilised as an excellent lateral stratigraphic marker, consisting of jaspilite and occasional hematite (Trendall & Blockley, 1970; Hamersley Exploration, 1972; Gilhome, 1975; Harmsworth et al., 1990). 13 Chapter 2: Geology and Hydrogeology 2.3.4 Mt. McRae Shale The Mt. McRae Shale (50 m thick) conformably overlies the Mt. Sylvia Formation. Its exposures are veiled by Brockman Iron Formation debris as a result of high weathering rates. This predominantly shale formation is made up of argillaceous materials of varying structure and colour according to the presence of free carbon. Pyrite nodules and zones of ferruginous concretions are abundant. These zones are not confined to individual beds, suggesting that they derived through chemical precipitation during diagenesis. There are also thin bands of volcanic shards in beds of volcanic breccia. (Trendall & Blockley, 1970; Harmsworth et al, 1990). The formation can be divided into four separate members according to lithology and pyrite content. The basal unit (15 m thick) is defined by a series of alternating carbonaceous shale and chert bands with increasing pyrite content at higher levels. Stratigraphically above this is a 10 to 15 m thick zone of alternating chert, black shale and minor doleritic shale (sometimes containing up to 7 % pyrite). Above the pyritic zone is a 10 m zone of non-pyritic which occurs as black shale when fresh (unweathered). The uppermost 12 m of the formation has been formalised as the Colonial Chert Member. This member comprises of thin BIF with interbedded shales (Trendall & Blockley, 1970; Harmsworth et al, 1990). 2.4 BROCKMAN IRON FORMATION The Brockman Iron Formation (~500-620 m thick) is the most economically important formation in the Hamersley Province due to its highly enriched BIF ore content. The unit varies in thicknesses ranging from 500 m in Paraburdoo and Newman to 620 m at Mt. Tom Price. The formation consists of alternating sequences of BIF, shale and chert. These can be divided into four primary members; the Dales Gorge Member, the Whaleback Shale Member, the Joffre Member and the Yandicoogina Shale Member (Harmsworth et al., 1990; Blake & Barley, 1992; Krapez, 1997). 2.4.1 Dales Gorge Member The Dales Gorge Member (~150-180 m thick) is an assemblage of 17 alternating BIF macrobands (DB0-16) and 16 shale macrobands (DS1-16) (Figure 2.1). These are divided into three units DG1-DG3: DG1 (to the base of shale band DS6); DG2 (base of DS6 to the 14 Chapter 2: Geology and Hydrogeology top of DS11); and DG3 (top of DS11 to the upper contact of the Member). These macrobands are generally laterally persistent throughout the region. The banded iron formations are usually made up of banded iron, chert, jaspilte, hematite and magnetite. The shale macrobands are usually unaltered, but can contain ferruginous zones of which iron percentages can reach up to 60% (Trendall & Blockley, 1970; Gilhome, 1975; Harmsworth et al., 1990). The Lower Ore Zone, DG1, consists of hematite, goethite, limonite with minor amounts of magnetite and quartz infill. Hematite ore in this zone can occur as massive, plately, friable or with ‘biscuity’ textures (Hamersley Exploration, 1972). 2.4.2 Whaleback Shale The Whaleback Shale (50 m thick) is a combination of two units, the lowermost unit comprising of four alternating macrobands of shale and BIF (WS1, WB1, WS2, and WB2). The uppermost unit WS1 is made up of numerous mesobands of chert and shale. WB2 is a 4 m thick cherty BIF and is typically crenulated (Gilhome, 1975; Harmsworth et al., 1990). WB2 is similar to WB1 and measures ~30 m in thickness (Hamersley Exploration 1972). 2.4.3 Joffre Member The Joffre Member (~360 m thick) consists of primarily BIF units with minor stilpnomelane- rich shale interbands and tuffaceous material. These minor interbands are dissimilar to the Dales Gorge member as they are thinner and not as laterally persistent. The member has been divided informally into six units named oldest to youngest as J1 - J6. Strands J1, J3 and J5 contain more shale than J2, J4 and J6 (Trendall & Blockley, 1970; Harmsworth et al., 1990). Banded iron is typically more abundant than the shale, especially where unaltered, and is made up of alternating bands of magnetite, hematite and chert. This typically alters to hematite, goethite, limonite and ferruginous shale (Hamersley Exploration, 1972). 2.4.4 Yandicoogina Shale Member The Yandicoogina Shale Member (~60 m thick) is composed of an alternating sequence of interbedded chert and shale. The western part of this member has been intruded by dolerite sills, as well as being locally enriched to form high grade hematite mineralization (Trendall & Blockley, 1970; Harmsworth et al., 1990). 15 Chapter 2: Geology and Hydrogeology 2.5 MT. TOM PRICE ORE BODIES The Mt. Tom Price area is made up of several discrete ore bodies occurring on or near the limbs of the Turner Syncline. The main ore body, the Mt. Tom Price deposit, measures approximately 7.5 km long and 1 km wide, with a pre-mining thickness of up to 250 m in certain areas. Mineralisation is also present at structurally prepared locations near the surface, as well as ‘deep resources’, which can be found up to 500 m depth below the pre-mining surface. Iron ore in this deposit is BIF derived with haematite dominance, and can be classified as Low P (Phosphorus) Brockman ore. The North Deposit (NTD), West Pits (WEST), Centre Pits (CTR), Southern Ridge (STR) and South East Prongs (SEP) pits are mining areas that are still producing this economical iron ore type (Figure 2.2). The South East Prongs deposit is the largest of these ore bodies, a 1.0 x 0.3 km synclinal structure with high grade mineralisation of the Dales Gorge member present to depths of 250 m. Mining areas to the south of the main Mt. Tom Price deposit are examples of High P Brockman ore (Haematite-goethite rich), that are typically capped with thick detrital and hydrated geothitic deposits. Section 7 (SSEV) measures 1.5 km long and 0.8 km wide, with mineralisation of the Dale Gorge Member occurring to depths of 120 m. Section 6 (SSIX), located east of this deposit is no longer operational, serving as storage for discharge water. Economical iron ore mineralization is also apparent along certain localities of the Marra Mamba Iron formation, south of the Mt. Tom Price deposit. The Marra Mamba East (MME) and Marra Mamba West (MMW) are future mining localities that will mine mineralized Marra Mamba ore up to depths of 200 m along two 4 km long pits (Figure 2.2). 2.6 BANDED IRON FORMATION (BIF) GENESIS AND CHARACTERISTICS 2.6.1 Supergene Enrichment Iron ore in the Mt. Tom Price area was formed by supergene enrichment, a process in which parent BIF is heavily altered to form high grade ore. This however, was selective and only limited amounts of BIF have been altered sufficiently to be mined economically. This was due to many contributing factors in its formation. These units had to be structurally prepared so that the process of groundwater percolation through BIF units could adequately increase 16 Chapter 2: Geology and Hydrogeology Figure 2.2: Geological map of the Mt. Tom Price mine area displaying geological structure and mining locations (Geological information from Rio Tinto Geological Database). 17 Chapter 2: Geology and Hydrogeology iron content. Units that are sufficiently folded and faulted are more favourable to enrichment, but also require sufficient climatic conditions, oxygen exposure, and suitable topography to create groundwater movement (Hamersley Iron, 2000). Supergene enrichment occurs through the part dissolution of gangue materials such as silica, as well the iron replacement of gangue through oxidation processes. This was selective as the units had to be structurally prepared. As a result, only a very small proportion of the total BIF in the region has become mineralized. Due to the dissolution and replacement of the gangue minerals, BIF units became up to 50-60% thinner than the parent BIF sequences (Hamersley Iron, 2000). Iron ore in the region occurs as three principal types: High P ore, Low P ore, and Marra Mamba ore. Low P iron ores formed about 1800-2000 million years ago through burial metamorphism and consequential dehydration. The goethite component of the parent BIF became dehydrated to microplaty haematite. This type of enrichment dominates in the Mt. Tom Price Mine area (Figure 1.6) (Hamersley Iron, 2000). High P and Marra Mamba ores were formed from similar enrichment processes some 60 million years ago. This process had a more widespread impact on the region, making up most of the mineralized deposits found in the region. As this was a fairly recent enrichment process, burial metamorphism was unable to influence enrichment, resulting in a haematite- goethite ore (Hamersley Iron, 2000). Parent material becomes heavily altered through oxidation, replacement of gangue minerals by iron oxides, and dissolution. Figure 2.3 illustrates the process of supergene enrichment, as summarised from a detailed CSIRO model (Hamersley Iron, 2000). 2.7 HYDROGEOLOGY OF MT. TOM PRICE 2.7.1 Groundwater Flow Groundwater flow in the Mt. Tom Price Mine area is relatively hydraulically continuous. Secondary structures such as folds, faults and dykes however, compartmentalize groundwater flow patterns in areas such as the South East Prongs, Centre Pit, West Pit, and the North Deposit. 18 Chapter 2: Geology and Hydrogeology Figure 2.3: Simplified model of supergene enrichment (Hamersley Iron, 2000) 19 Chapter 2: Geology and Hydrogeology The hydrogeology of the Mt. Tom Price Mine area is somewhat complex due to the structural setting of the region and the presence of secondary structures such as dykes and faults. Regional groundwater flow has historically been defined as south-westerly, with minor radial flow occurring between NTD and SEP due to groundwater compartmentalisation (Preston, 1994). This perched body has been included into the model and water level contours to confirm that this is in fact a perched body, rather than a mound as suggested by Preston in 1994. An anticlinal structure with high permeability contrasts acts as a barrier to flow between these areas. Impermeable dolerite dykes and fault gouge associated with the Southern Batter Fault are likely to aid groundwater compartmentalization. Pumping which has occurred since 2001 has had little effect on the perched groundwater mound, suggesting that this is likely to be a perched aquifer. Another possibility is that current pumping regimes are not hydraulically connected, resulting in little change to the groundwater mound post- pumping. Figure 2.4 illustrates earliest known piezometer records (1994) and the June 2007 water level contours. Geologically, the area is made up of impermeable BIF, shales and chert formations. Due to Supergene enrichment, units such as the Dales Gorge Member, Footwall Zone, Joffre Members and the Marra Mamba Iron Formation can become many times more permeable, acting as highly transmissive aquifers where the units are structurally prepared and mineralised. Where mineralised, the units become much more friable, and adopt a ‘biscuity’ texture that allows water to pass through at hydraulic conductivity rates of 1-5 m/day (pers. comm. Rathbone, 2007). Other units of high permeability include the Bruno’s Band, located in the Mt. Sylvia Shale Formation. This unit is typically composed of impermeable shales, with a 10 m chert band that can act as an aquifer, transmitting water at 1–5 m/day where the unit is faulted and folded (pers. comm., S. Rathbone, 2008). Previous mining over the last 40+ years and future mining to come should effectively leave the area stripped of most of the highly permeable units. The few highly permeable areas remaining will be associated with faulted units and fault structures where fault gouge does not impede groundwater flow. The Bruno’s Band aquifer should also remain relatively unaltered. 20 Chapter 2: Geology and Hydrogeology a) Feb-May 1994 b) June 2007 Figure 2.4: a) Feb-May 1994 water levels and Tom Price geology. b) June 2007 water levels and Tom Price geology. NB Contour readings are in mRL (mine reference level) values against the Tom Price mine grid. It is important to note that areas outside the mine area should be considered subjective due to the lack of observation wells outside of the individual mines. The perched water body in the centre of the plot should be ignored. 21 Chapter 2: Geology and Hydrogeology Highly weathered zones and detritus associated with erosion and down cutting by creeks may have a minor effect on groundwater flow throughout the mine area. Most of these zones are not in direct hydraulic contact with the generally deep water table and therefore are not a primary focus in this study. Secondary permeability changes associated with mining activity can also give unmineralised BIF higher permeability, putting it in direct hydraulic contact with the water table. This zone could perhaps be applied to a zone 10 m wide behind and below the pit walls and floors of spent mines, due to intense, blast induced fracturing (pers. comm., S. Rathbone, 2007). Increased permeability may also result from unloading by the removal of overlying material during mining procedures (pers. comm., G. Domahidy, 2008). Dolerite dykes that have intruded away from fault zones can have very low orders of conductivity values, acting as a ‘wall’ or dam against groundwater flow. Groundwater action against the dolerite surface weathers it to impermeable clays, with hydraulic conductivities of the order of 0.00001 m/day. The contact with the surrounding BIF units is often permeable, especially around areas of faulting. Here, conductivity values can be of around 1 m/day, acting as a conduit for groundwater flow against an impermeable barrier (pers. comm., L. Campbell, 2007). A number of faults run through the mining areas and these often have a dramatic effect on groundwater flow. The Southern Batter Fault displays up to 100–150 m displacement in some areas, positioning the more permeable basement rocks against the less permeable up- sequence units, compartmentalizing groundwater flow (Figure 1.6). Along certain lengths of this fault line, a distinct zone made up of highly faulted BIF and shale acts as a conduit for groundwater flow (pers. comm., L. Campbell, 2007). Smaller, highly permeable fault zones also occur along the length of the fault. As a result, a localised step down in water levels can be observed in these areas. In some areas the fault zone has developed a fault ‘gouge’, consisting of pulverized clay sized sediments of BIF and shales. This impedes water flow through the fault face, further aiding to compartmentalization. 2.7.2 Dewatering History Large scale dewatering programs commenced early in 1994, when pit floor progression encountered regional and perched water tables for the first time. North Deposit was the first 22 Chapter 2: Geology and Hydrogeology mining area that required large scale pumping, with a single pump (DB4) extracting approximately 300 kL/day from 1994-2004. Bores WB03NTD1, WB05NTD1, WD06NTD1 and WD06NTD2 have since been installed, pumping volumes of between 300–2500 kL/day. South East Prongs was also an area of intense pumping once dewatering commenced, with bores DB1-3 pumping some 300 kL/day. Currently there are 3-4 active pumps (WB05SEP1, WB05SEP2, WB06SEP1, and WB06SEP2) operational at any one time, each pumping volumes between 420–5160 kL/on average. Section Six had one pump operating from 1994-2003, now known as WB05SSIX1, extracting on average 490 kL/day. The Southern Ridge pit is host to a single bore, WB04STR2, pumping sequentially according to required volumes, ranging from 170–1050 kL/day. Section Seven has just one pump operating, WB05SEV01, extracting 490 kL/day on average A complete record of pump locations is included in Appendix 19 (results approximated from Rio Tinto Hydrogeological Database, 2008). 2.7.3 Piezometer & Monitoring Bore Network Since 1993, various drilling programs have been implemented to provide coverage of groundwater levels within the Mt. Tom Price mining area. This network has been established to ensure that water levels are sufficiently reduced for pit level progression and future mine planning. Regular readings are also useful to observe the dewatering effect of the abstractions, as well as to provide information on zones of influence, such as cones of depression, depressurization requirements, and other geotechnical purposes. In order to observe rock mass potentiometric levels (the water ‘head’), a series of piezometers and open standpipes have been installed. Piezometers in the mining area screen water bearing units such as the Dales Gorge Member, Mt. McRae Shale, Mt. Sylvia Formation, and the Wittenoom Formation. These piezometers range in depth from 12 – 180 m with variable screen depths and intervals depending on hydrogeological structure. To date, there are 62 operational piezometers and standpipes in the mining areas, of which several can become inaccessible at any time due to mining activities. Figure 2.5 displays operational bores and 23 Chapter 2: Geology and Hydrogeology piezometers in the mining area as of June 2007. Complete observation records can be found in Appendix 19. Piezometers and water bores are designed to be placed in areas where the effects on the mining process are minimal. However, due to changes in pit designs and mine expansion, many of the previously installed piezometers and bores have been mined out or access has been restricted. Readings are taken on a weekly basis by onsite technicians and hydrogeologists, as well as monthly water sampling by the environmental team. Measurements are taken using electric water level probes and recorded directly into a excel spreadsheet template that comprises part of the Rio Tinto Hydrogeological Database. 2.7.4 Dewatering Target Areas Since 1994, mine hydrogeologists have planned dewatering programs according to final mine plans and progressions. Table 2.1 and 2.2 display a 1994 dewatering program plans, addressing groundwater mRL requirements with pit floor progression. From this report, the North Deposit and South East Prongs are the main areas of concern, requiring dewatering of up to 1800 ML, producing a drawdown of approximately 100 m (Preston, 1994). Dewatering requirements depend upon how much water is in storage that needs to be depleted, leakage from other units and the mining rate (pers. comm., G. Domahidy, 2008). As final pit design deepens, dewatering requirements increase. Prior to mine closure, the following areas will need dewatering, ranked in order of pumping requirements. 1) South East Prongs 2) North Deposit 3) Marra Mamba 4) West Pit 5) Centre Pit (perched aquifer system) 6) Section Seven (perched aquifer system) As of January 2008 there are 7 active dewatering bores, abstracting primarily from the regional groundwater table in the SEP and NTD mine areas. Each bore pumps an average of approximately 2000 kL of water daily. Final pit plans suggest that pit floor in the Section 24 Chapter 2: Geology and Hydrogeology Figure 2.5: Mt. Tom Price water bore and current (June 2007) piezometer locations (Hart, 2007 – Rio Tinto Hydrogeology Database) 25 Chapter 2: Geology and Hydrogeology Seven pit will be above the regional water table, however, it is likely that perched aquifer systems may require continued pumping prior to closure. Table 2.1: Historic dewatering targets (mRL) (Preston, 1994) South East South East End of South East North Section Prongs Prongs West Pit Mining Year Prongs Deposit Six Extension South SEP SEPX SEPS WPIT NTD SEC6 1995 675 + 690 750 + 705 1996 660* + 660 735 + 705 1997 645* + 660 + 735 690 1998 645* + - + 735 690 1999 645* + - + 720 675 2000 - 675 - + 720 675 2001 - 645 - + 705 675 2002 - 645 - 735* 705 675 2003 - 630 - 720* 660* 660 2004 - 615 - 720 660* 630* 2005 - 600 - 849 645* - 2006 - 585 - 836 630* - Approx. Groundwater 660 660 660 760 670 645 Level * = below present groundwater level Table 2.2: Historic Dewatering volumes (Preston, 1994) South East West Pit North Deposit Section Six Prongs SEP, SEPX, WPIT NTD SEC6 SEPS Dewatering 600-690 ML 150-400 ML 1300 ML 125 ML Requirement Indicated Wall- From below 660 Rock Drainage/ mRL from the From below 650 Perhaps needed Not needed Depressurization FWZ and the Mt. mRL from shales requirements McRae Shale (? Horizontal (? Horizontal drains) drains) 2.7.5 Groundwater Management Groundwater extracted from the numerous pumping bores is collected and transmitted in a pipe network that leads to storage tanks, processing plants, or as environmental discharge. Alternatively, groundwater is pumped away from active dewatering areas into worked out mine sites, such as Section Six, where approximately 1000 kL are deposited daily on average (June 2007). Groundwater is also used for dust suppression on mining areas and haul roads. 26 Chapter 2: Geology and Hydrogeology A 2007 dewatering plan is included in Appendix 1 in order to summarise dewatering networks. Although most elements in this plan are outside of the area modelled in this project, it is helpful to visualise the current (2007) dewatering storage areas. 2.7.6 Slope Depressurization In response to mine floor progression, slope depressurization is an area of concern in the numerous mining areas. Water table levels within and around the pits are important, as are surface water drainage procedures. Poorly drained pit walls can increase pore water pressures, resulting in significant factor of safety decreases. In order to depressurize the pit walls, horizontal drains have been installed along various pit walls (Figure 2.6). The result is a drawdown in the zone of influence of the drains, which often produces both a depressurization and dewatering effect (Preston, 1994). Current plans involve the installation of approximately 130 drains in the SEP and NTD at drill depths of between 50–100 m at 25 m offsets (Rio Tinto Hydrogeological Database, 2007). Figure 2.6: Section of horizontal drain holes in pit wall (Hydrogeology Concept Projects - Rio Tinto Hydrogeological Database, 2007). 2.7.7 Surface Water Management Precipitation from normal seasonal rainfall events in general has little effect on long term water level fluctuations. Due to the high evaporation rates in the region, only around 1% of 27 Chapter 2: Geology and Hydrogeology recharge reaches the groundwater table (pers. comm., K. Rozlapa, 2007). Surface ponding and runoff however, are likely to contribute to water table variances. This occurs typically through broken rock in the pit floors, as well as along faulted and fractured rock zones throughout the area. Intense, irregular (extreme) storm events are also likely to have an influence on water table fluctuations. During these events, a Probable Maximum Precipitation (PMP) of 440 mm can fall over the period of one hour. It has been determined that there is a 1:20 event probability in the Mt. Tom Price mining area for runoff and ponding to occur. All rainfall on slopes can be considered runoff, due to the lack of vegetation and rocky slopes (Preston, 1994). The large open pits in the Mt. Tom Price mining area are situated near topographic divides, effectively acting as their own micro-catchments. Extreme rainfall events, as well as above average periods of rainfall, can lead to significant ponding in berms on the pit walls and surface erosion of pit slopes and haul roads. Local creeks will also flow during these events, and it is therefore necessary that flow into the pits is diverted to limit flooding. Ponding within these pits can eventually infiltrate through the pit floor, creating perched water tables on previously dewatered rock units (Preston, 1994). 2.8 SYNTHESIS The geology of the Mt. Tom Price area comprises of a series of banded iron formations and shales that are generally low in hydraulic conductivity values. The process of supergene enrichment involves a dissolution process whereby gangue minerals are dissolved and replaced with iron variants such as haematite and goethite. This process has a tendency to increase hydraulic conductivity values by several orders of magnitude, resulting in high yielding aquifers within the pit boundaries. Highly faulted and folded units can also have augmented hydraulic conductivity values. Since 1993 a series of observation bores have been installed in the mine area. Pit floor lowering began to encounter the regional water table at this time. It is essential that dry mining continues, therefore a series of dewatering bores a depressurization measurements have been implemented. The data collected from dewatering and observation bores can be used to help understand regional groundwater flow and create the MTPGM. 28 Chapter 3: Geological & Conceptual Model CHAPTER 3 GEOLOGICAL MODEL & CONCEPTUAL GROUNDWATER MODEL 3.1 INTRODUCTION A 3D geological model of the Mt. Tom Price mining area was constructed from existing drillhole databases, geological cross sections, and Vulcan triangulation models so that geological units and structure were captured correctly. These databases were used in conjunction to ensure that accurate representations of the key geological features in the mining area were imported into the MTPGM. Capturing the geology accurately is essential when constructing a groundwater flow model. An accurate 3D model will aid in the visualisation and understanding of the geological structure of the Mt. Tom Price area. This has allowed for the creation of detailed layer template files that can be entered directly into the MTPGM. 3.2 RIO TINTO DRILLHOLE DATABASE The data contained within the Rio Tinto Drillhole Database has been collected since exploration began at Mt. Tom Price in the 1960s. Generally, each drillhole has been accurately documented in the database. These records contain northing and easting coordinates, geological unit interpretations, chemistry, and Iron (Fe) content for most drillholes in all of Rio Tinto’s mining operations (Figure 3.1). Records can be exported to various formats, such as an excel file where drillholes can be sorted and manipulated as required. As there are literally thousands of drill holes to choose from, 22 cross sections were selected that had optimal north-south coverage over the entire mining area. These have been labelled CS1-21 and CS25, and are base points for continuity between the geological datasets (Figure 3.2). Three east-west cross sections (CS22-24) were also created for continuity, but were not 29 Chapter 3: Geological & Conceptual Model drafted. A complete collection of the cross sections with geological interpretation data (which will be discussed in section 3.7) have been included as Appendix 2. Figure 3.1: Aerial view displaying complete Mt. Tom Price drillhole database. Red circles represent available data. The Yellow boundary represents the catchment boundary and project area. For this geological model, simple strand interpretations of the different geological units were required. These were then assigned x, y, z values in order to create a 3D representation of the depths and thicknesses of the geological units throughout the mining area. These were exported to .dxf files using TERRAMODEL (v.10.41, Trimble, 2005). This data was then imported into AUTOCAD 2002 (v.15.6, Autodesk, 2002) where the strand data was linked using polylines to create accurate geological cross sections (Figure 3.3). 3.3 HISTORICAL GEOLOGICAL CROSS SECTIONS In 1972 the Mt. Tom Price mining area was geologically mapped in extensive detail to a scale of 1:4800 (Figure 3.4). Modern geological data is focused in and around pit localities. This 1972 Hamersley Exploration study is the only available report on regional geological units at depths away from the pits. The exercise was undertaken in order to improve structural and stratigraphic interpretations, and to define mineralisation boundaries more precisely. Regularly spaced cross sections were selected according to drillhole location availability (Figure 3.5). The modern datasets such as drill hole database interpretations and Vulcan 30 Chapter 3: Geological & Conceptual Model 2.3 for geological to Figure Refer s. ppendix\x 2 for full records). 2 ppendix\x cross-section Cross-section lines (CS) represented by red lines (Refer to A lines (Refer by red represented (CS) lines Cross-section ng sampled drillholes (red) and their corresponding key. NB Areas coloured black represent remaining mineralised BIF. mineralised remaining represent black coloured Areas key. NB 31 displayi 3.2: Geological Map of the Tom Price mine area Figure Chapter 3: Geological & Conceptual Model Figure 3.3: Example of interpreted cross section drillhole values entered into AUTOCAD (looking west through CS1). Figure 3.4: Map of cross section localities from 1972 1:4800 survey. Coordinates are displayed in the old Tom Price mine grid format (Hamersley Exploration, 1972). Figure 3.5: Cross section example through 1800 E (15300 E) (looking east) (Hamersley Exploration, 1972). 32 Chapter 3: Geological & Conceptual Model triangulation models were combined for improved accuracy. These datasets generally proved to be accurate in and around the actual pits, with accuracy decreasing away from the pits due to reduced drillhole frequency. This report has been included as Appendix 22. 3.4 VULCAN TRIANGULATION SURFACES Vulcan triangulation surfaces have been constructed for mine planning, geotechnical and hydrogeological purposes. These models have been developed through the process of polygonal triangulation using data from the Rio Tinto Drillhole Database. They act as best- estimate surfaces of sub-units in the mining area, including: Joffre (J1-6), Whaleback Shale (WBS1-2), Dales Gorge member (DG1-3), and the Footwall Zone (FWZ). Surfaces can be viewed in 3D visualisation programs such as Vulcan or AUTOCAD in order to aid in mine planning procedures. The area outlined in Figure 3.6 displays reliable data created from intense drillhole programs and can be considered accurate in and around the pit localities. Raw drillhole data was therefore not required for correlation in these areas. 3.5 CONSTRUCTION OF GEOLOGICAL MODEL Using a 3D visualisation program (AUTOCAD), the drillhole strand interpretation values can be imported into an environment referenced to the Tom Price Mining Grid (TPMG). Cross sections can then be drafted so that accurate representations of the geology can be visualised. This also helps to identify folds and faults that may have been missed by previous drilling programmes and geological assessments. An overhead view of the 3D geological sample lines is included as Figure 3.7. The complete model has been included as Appendix 20. As all of the units have been converted to the Tom Price mine grid, the cross sections, drillhole data and Vulcan surfaces can be simply entered into the same modelling environment. These can be visually checked for consistency and adjusted accordingly. To aid visualisation, units can be ‘hatched’, so that they appear as solid, filled units. This method could be applied to any desired unit or area. In this model, the Dales Gorge Member and the Footwall Zone (potential aquifers) have been hatched with a light blue design, while the Marra Mamba Iron Formation has been hatched with an aqua design. 33 Chapter 3: Geological & Conceptual Model SEP STR Figure 3.6: Oblique 3D view displaying a sample of Vulcan triangulation surfaces (mining locations highlighted in red). Figure 3.7: Geological Model (map view). Vulcan surfaces (pink) used where cross sections are unavailable. 34 Chapter 3: Geological & Conceptual Model In AUTOCAD, or a similar 3D modelling application, the cross sections can be rotated so that structure becomes apparent. This is useful to understand the structure of the regional and secondary synclinal/anticlinal structures in the Mt. Tom Price mine area. An export has been included as Figure 3.8, where clear geological structures such as synclines, anticlines and faults can be easily identified. 3.6 CONCEPTUAL GROUNDWATER MODEL The relationship between the geological and the hydrogeological formations in the Mt. Tom Price area is one that is relatively well understood but poorly documented. Mineralised banded iron formations and chert layers make up the major hydrogeological units in the mining area. As the main objective of this thesis is to assess groundwater levels and conditions upon mine closure, mineralised material can effectively be negated from the MTPGM final predictions, simply as the majority of this material will be extracted from the area. Therefore, bedrock groundwater flow through tight, unmineralised BIF and shales dominates in the mining area. The conceptual MTPGM was based upon the various geological data, including the 1972 geological survey (Hamersley Exploration Pty Ltd., 1972), as well as Vulcan triangulation surfaces generated from resource exploration and definition drilling. As the modelled area was relatively large (~11 x 11 km), small scale structures such as faults and folds were not represented in the groundwater simulation. However, where large displacements occur, lithological changes across a layer should represent the faults sufficiently, given the model scale. The large Southern Batter Fault zone is represented in the model as a 40–80 m wide, sub-vertical, conductive structure running E-W for approximately 1 km. Hydraulic head values were obtained from the Rio Tinto Hydrogeological Database. This data is collected from weekly monitoring of the many piezometers within the mining area. Pumping values were also obtained from the Rio Tinto Hydrogeological Database as collected from historic and weekly monitoring data. Due to the highly variable topography and nature of the mining area, specified reference level values have been selected to define the model layer elevations. The MTPGM covers the area within the catchment boundary defined by elevations ranging from 820 mRL to 520 mRL. 35 Chapter 3: Geological & Conceptual Model en, Faults en, = Red, Mineralised = Gre Dykes (light blue/aqua). FWZ/Marra Mamba units highlighted Mamba units highlighted FWZ/Marra BIF = Black. Figure 3.8: Oblique 3D view of geological model. Dales Gorge + + Gorge Dales model. geological of view 3D Oblique 3.8: Figure 36 Chapter 3: Geological & Conceptual Model These elevations were chosen to incorporate the maximum height of possible water level recovery at the highest elevation, while capturing the lowest mining level and the surrounding geological structures. Elevations were also selected to visualise the perched groundwater body in the Centre Pit-Southern Ridge area. Preston suggested that this perched body was a groundwater mound, therefore it should be helpful to enter this into the initial MTPGM to see if this is valid (Preston 1994). Initial MTPGM results should indicate that this mound is indeed an unresponsive perched body, and can be negated from simulations if need be. The MTPGM only simulates saturated flow, so if the high groundwater elevations observed are proven to be a perched body, it should be ignored when calibrating to measured data as this will not be accurately represented. This is because the model has set up for saturated basement flow and will therefore not accurately represent unsaturated material upon simulation. An initial water balance estimate was constructed as follows. Recharge values were obtained by averaging historic rainfall data (~300 mm/year) to calculate a representative daily recharge value. Due to high evaporation rates in the region, only ~1% of this recharge will infiltrate down to the water table and influence head changes. This was used to simplify the MTPGM and allow for a more practical basement flow groundwater model simulation. Obviously this method will result in simplified results, but was considered acceptable for this project. Hence, a very low recharge value has been assigned to the model. Total water balance input and output values were evaluated from initial runs of the MTPGM (as discussed in Chapter 4). Results indicated that approximately 200 ML of water recharges the aquifer per year. 1000 ML is extracted by pumping, and 180 ML flows out of the model basins at the fixed head boundaries. Pumping and discharge values were also acquired from the Rio Tinto Hydrogeological Database. These values represent averaged values of modern day (2007) pumping and discharge regimes. Water moving in and out of storage was calculated from initial runs of the MODFLOW model, which will be covered in Chapter 4. The conceptual groundwater model included as Figure 3.9 gives a good overview of the hydrogeological parameters and stresses throughout the entire modelled area. Initial parameter values were collected from lab testing and previous hydrogeological work (Appendix 3). 37 Chapter 3: Geological & Conceptual Model runs. model DFLOW 07 available data and initial MO available data and initial 07 ogical interest. Taken from June 20 Taken from interest. ogical of hydrogeol main areas M through 38 Figure 3.9: Conceptual MTPG Chapter 3: Geological & Conceptual Model 3.7 RELATIONSHIP BETWEEN CONCEPTUAL AND NUMERICAL MODELS The relationship between the conceptual and a numerical groundwater model is important in understanding how the hydrogeology of a modelled area is numerically represented in a groundwater model. As detailed 3D geological information can be easily transferred into the modelling environment, the MTPGM can be considered to be a direct representation of the geological model. The MTPGM is a uniform grid, with 12 flat-lying layers that each represents many different hydrogeological units within a single layer. To make sure these units are represented accurately in each layer, the geological data from the geological model is used as the basis for entering the various hydrogeological parameters into the model. Figure 3.10 shows the grid setup against the conceptual/geological model. This setup is also useful in visualising the detail lost in groundwater models. It can be seen that the thinner units are not-hydraulically connected in some layers (Figure 3.10). This could have been fixed by increasing resolution, but would have resulted in impractical model run times. To ensure that the hydrogeology is accurately represented, the different geological units need to be defined according to the planned layers in the hydrogeological model. 6 layers can be sampled at 50 m increments in order to provide coverage of all the pit voids and surrounding geology. This simplification was used due to time constraints involved with the creation of the geological model. Each cross section should be sampled individually. Coloured 50 m high rectangles can be inserted to represent the different units, eg. Joffre = blue, Dales Gorge = brown. This data can then be combined to show a better representation of geological structure. The geological data combined with cross sections 1-25 are included in Appendix 2. Figure 3.11 shows an oblique 3D veiw of the enitre interpreted geological model in the Mt. Tom Price Mining area. This model has been included in Appendix 21. From this regional model, each 50 m layer can be individually selected and entered into a new environment. From here, the gaps within the model can be filled in through interpretation between the coloured rectangles. Polylines can be created and hatched according to different geological units. Once interpreted, these files can be saved in .dxf format so that they can be opened in PMWIN Pro. This process was repeated for each layer in the model, resulting in a complete geological template (Figure 3.12) to model the- 39 Chapter 3: Geological & Conceptual Model s (dark grey). grey). (dark s l re 3.9 for key). key). for re 3.9 gical to units (Refer Figu hydrogeolo different splaying al cross-section through 15720 E (looking east), di east), (looking 15720 E through al cross-section A B B) Cross-section through the Mt Tom Price groundwater model at 15720 E. Unsaturated rock mass has been assigned as inactive cel inactive as assigned been has mass rock Unsaturated E. 15720 at model groundwater Price Tom Mt the through Cross-section B) Fig 3.10: A) Conceptu 40 Chapter 3: Geological & Conceptual Model Figure 3.11: Oblique 3D view in the modelling environment (looking north-west) displaying Interpreted geology. Geological units assigned indiviual colours. Refer to Appendix 2 for key. Figure 3.12: Sample of geological template file between 720-670 mRL in map view (Layer 7-8). Refer to Appendix 4 for key). 41 Chapter 3: Geological & Conceptual Model hydrogeology of the Tom Price mining area. The complete 6 layers covering the entire Mt. Tom Price mine area are presented in Appendix 4. Where increased detail is needed, separate layers can be created at a 25 m resolution in order to capture thinner geological units or pick up hydrogeological detail in and around the pit- voids. Using the .dxf template files, parameter values can be assigned using the polygon tool in PMWIN. These polygons can be saved and reused with different parameter values, speeding up the input process significantly. 3.8 SYNTHESIS The creation of an accurate 3D geological model is essential for visualisation of semi- regional geology and accurate representation of hydrogeological structures. A geological model was constructed from drillhole records, historical records and Vulcan triangulation models. From this geological model, accurate template files were created so that geological detail loss is kept to a minimal when entering hydrogeological parameters into the MTPGM. To help with model setup, a conceptual groundwater model was created using information from the geological model, field tested hydrogeological parameter values and field sampled measurements. 42 Chapter 4: Mt. Tom Price Groundwater Model CHAPTER 4 MT. TOM PRICE GROUNDWATER MODEL 4.1 INTRODUCTION The semi regional hydrogeology of the Mt. Tom Price mine area is poorly understood. Hydrogeological work in the area focuses on dewatering and depressurization requirements to ensure dry mining conditions, rather than understanding the connectivity, if any, between the individual pits and mining areas. Groundwater flow patterns in the Mt. Tom Price area can be considered as a bedrock aquifer system, with flow through discontinuities within the tight BIF and shale units. Compartmentalisation of flow patterns occurs as a consequence of the complex geological structure of the area, resulting in a number of perched water bodies and aquifer systems in the mine area. Mineralisation and fault zones can act as conduits for groundwater flow, due to permeability augmentation upon formation. Using a 3D modelling approach, bedrock aquifer flow can be comprehensively modelled according to field measurements and hydrogeological parameters taking into account connections between neighbouring mining areas. Using this model, it is possible to predict groundwater flow and final pit water levels so that mine closure procedures can be planned and implemented. 4.2 PREVIOUS GROUNDWATER MODELS A number of groundwater models have been constructed in the Tom Price mining area by the consulting company Aquaterra. These models have primarily focused on dewatering requirements of individual pits, including the South East Prongs, North Deposit and Section Six, as well as water quality estimations from Southern Ridge and South East Prongs. These models are helpful as they provide data and results that will aid in the setup of the MTPGM 43 Chapter 4: Mt. Tom Price Groundwater Model as discussed in section 4.3. Parameter Values used in the previous models should be similar to those used in the MTPGM, and final outputs from these models can be taken into consideration. All previous model reports discussed have been included as Appendix 23. 4.2.1 South East Prongs Dewatering Model (Ariyaratnam and Hall - Aquaterra, 2001) In 2001, Pilbara Iron (then known as Hamersley Iron) commissioned Aquaterra, a hydrogeology consultancy, to construct a groundwater model to be used in determining the dewatering requirements for a new pit design. The model was created using the PMWIN Pro graphical interface, and consisted of a two layer system. The mineralized Dales Gorge Member and the Mt. McRae Shale were represented as two separate, uniform layers. Cell size was set to 10 x 10 m uniform cells covering 1.5 km2 of the South East Prongs pit boundary. Layers followed the base of the Mt. McRae Shale, according to mRL values from drillhole data. The model was bounded on all sides with no flow boundaries, with no set background recharge or evaporation values. Parameter values entered into the model were as follows: Table 4.1: Parameter values for the SEP dewatering model (Ariyaratnam and Hall, 2001) Parameter Layer 1 (Mineralized DG) Layer 2 (Mt. McRaes Shale) Aquifer Type Unconfined Confined/Unconfined -3 Hydraulic Conductivity (Kh = Kv) 20 m/d 1 x 10 m/d Specific Yield 0.1 5 x 10-3 Storage Coefficient - 5 x 10-5 This model is essentially a groundwater storage depletion model, which can be used to determine dewatering requirements upon pit redesign. In 2004 this model was used to assess the dewatering requirements of the bores DB3 and WB03SEP-01 with varied recharge scenarios. Water levels simulated using no recharge remained steady due to the use of no flow boundaries. Simulations with high recharge, similar to the volumes measured during the high rainfall of the 2004 wet season (November – April), showed that current pumping rates (~1.5 ML/day) would be insufficient to continue with dry mining. The model also indicated that continuous pumping at 1.5-2.0 ML/day for 12–18 months should be sufficient to dewater the ore body aquifer. Figures 4.1-4.2 illustrates the need for increased pumping relative to pit 44 Chapter 4: Mt. Tom Price Groundwater Model floor progression. Figure 4.1 shows that as at 1999, pumping levels needed to be dramatically increased so that dry mining practices could continue. Figure 4.2 displays predicted water levels below 630 mRL with 2001 pumping rates. Figure 4.1: Final head values at last stress period (2014) with 1999 pumping abstraction rates. Black line represents final pit floor levels (Hall, 2001). 1 1 1 Figure 4.2: Areas of predicted groundwater levels below 630 mRL with 2001 pumping rates (Oct 2004) Note: Pit toe line 630mRL shown (Red line). Dates are only approximate. Outer margins of ore body are currently desaturated (Rozlapa, 2004) 4.2.2 North Deposit Dewatering Model (Rozlapa & Hall, 1999 (096a.pdf)) In 1999, a model similar to the South East Prongs model was constructed by Aquaterra to determine dewatering requirements for future mining plans of the North Deposit. This model was constructed to assess the performance of current and proposed bores south of the pit. This report concluded that with all 4 bores operational (NDDW1 + 3 proposed hypothetical bores) pumping at 1.5 ML/day (6 ML/day total) was estimated to be sufficient until September 2006. Figure 4.3 illustrates dewatering requirements by displaying target water levels against predicted water levels, suggesting that new pumps would need to be installed 45 Chapter 4: Mt. Tom Price Groundwater Model with combined pumping rates of 3.3-3.4 ML/day from January 2003 onwards. These results were implemented into plans and reflect actual dewatering requirements. As mining progressed, there was a slight increase in these predicted values, but the model was determined to be essentially successful. As a September 2006, there were 3 bores operational, pumping 2.3 ML/day, 1 ML/day and 1.2 ML/day (Rio Tinto Hydrogeological Database, 2008). Figure 4.3: North Deposit dewatering requirements displaying simulated waterlevels vs. pit floor progression (Rozlapa 1999). 4.2.3 Section Six Storm Water Storage Model (Druzynski & Hall, 2004 (R075b.pdf)) In 2004, in response to intense short-period rainfall events, Aquaterra were commissioned to assess the storm water storage potential of the Section Six pit for ponded surface water from the South East Prongs pit. The model was constructed in a similar fashion to the previous Aquaterra models; however 4 layers were used to simulate the pit void, Dales Gorge Member, Footwall zone, and Mt. McRae Shale. Fixed head boundaries were assigned to the north and south boundaries, with 2.5 m/year evapotranspiration rates applied to layer 1. The following parameters were entered into the model (Table 4.2), calculated from lab and field testing. These parameter values are higher than expected, likely a result of model calibration so that outputs matched measured values. 46 Chapter 4: Mt. Tom Price Groundwater Model Table 4.2: Model parameters for the SSIX storage model (Druzynski & Hall, 2004). A steady state and transient calibration was applied to the model, with values compared against records over the period between May–September 2004. The model was then run to assess the potential impacts of future pit discharges, using the MODFLOW particle tracking package, specifically during an intense rainfall event. The results indicate that after 1 year, seepage would reach 100 m from the pit, moving rapidly due to the steep hydraulic gradient from the pit lake and surrounding water table. After 10 years, seepage would reach 300 m from the pit, and finally 1200 m away after 1000 years (Figure 4.4). This flow would dominantly move west through the tight Dales Gorge member, having little impact on groundwater resources in the region. Figure 4.4: Section Six Particle travel over 1000 years following discharge of excess water from intense 1:20 year flooding (160 ML) (Druzynski & Hall, 2004). 47 Chapter 4: Mt. Tom Price Groundwater Model 4.2.4 Pit Void Closure Modelling (Rozlapa & Hall – Aquaterra, 2002) In April 2002, using the NTD and SEP models available, Aquaterra undertook an investigation on the potential of final water levels in the SEP, NTD and Southern Ridge pit voids. It was apparent from this report that further investigation was required to obtain more reliable results. This thesis attempts to address the unknown SEP conditions highlighted in this report. The results indicated that the North Deposit will likely fully refill to 662 mRL after 40 years (Table 4.3). Southern Ridge estimates were unable to be calculated as models were unavailable. Interestingly, the South East Prongs pit is expected to be dry, which seems unlikely as final pit designs are approximately 100 m below the original pre pumping water table. Table 4.3: Predicted long term pit lake conditions at Mt. Tom Price (Rozlapa & Hall, 2002) Pit Predicted Long Term Conditions Predicted Pit Time Taken for Time Taken For Full Pit Lake Volume Void Water 50% Recovery Recovery (Years) (ML) Level (mRL) (Years) 662 40 years 6,600 North Deposit Dry pit expected NA NA NA South East Prongs ? ? ? Southern Ridge 4.3 MTPGM MODEL SETUP A transient groundwater model was created using MODFLOW 2000 (v.1.15, Harbaugh, 2005). The graphical user interface Processing Modflow Pro (PMWIN Pro) (v.7.0.34, Chiang, 2006) was used with MODFLOW, using data from the geological model as described in Chapter 3. PMWIN acts as a 3D graphical user interface to the MODFLOW program, allowing users to enter hydrogeological data directly into a 3D environment. The geological structure in the Mt. Tom Price mining area is generally steeply dipping and highly folded. When modelling dipping strata in MODFLOW, errors can occur depending on the level of dip angles. Geological units dipping above ~30° can become problematic when modelling individual geological units as separate layers. At these angles, gravity driven flow begins to dominate over horizontal flow. To overcome this, a uniform grid should be used so 48 Chapter 4: Mt. Tom Price Groundwater Model that each cell over the entire model can be assigned discrete parameter values. However, this method results in a greatly increased construction time for the model, as the hydrogeological parameters must be appropriately mapped out for each specified layer. Detail of the connection of the units is paramount as units need to be touching in the x, y, and z directions to allow for water transmissivity through a unit (pers. comm. Rozlapa, 2007). 4.3.1 Model Input Data 4.3.1.1 Initial Model Setup To set up the model, a grid size of 50 m by 50 m was defined with 264 columns and 218 rows. The environment was created so as to be geographically referenced to the Tom Price Mine Grid. This can be edited in the ‘Environment’ menu, where the following coordinates were defined (Figure 4.5): Figure 4.5: MTPGM coordinate system values for use in PMWIN Pro. 4.3.1.2 Boundary Conditions The MODFLOW flow model requires an IBOUND array, which contains a code for each model cell. Values can be assigned according to whether a cell will be active (a positive value), whether a cell will remain inactive (a zero value), or whether a cell will be given a fixed head (a negative value) (Chiang, 2005). Active cells are water bearing areas of the model that lie within the specified catchment boundary, defined from local topography. Hydraulic heads within these cells are computed 49 Chapter 4: Mt. Tom Price Groundwater Model and are usually given a value of ‘1’ using the IBOUND array. Inactive cells are areas that lie outside the catchment boundary and or are permanently non water bearing. These cells are assigned a value of ‘0’ in the IBOUND array. These are coloured grey so that no flow is allowed to take place in the cell (Figure 4.6). Areas within the model that are known to have a constant head can be specified as a fixed head cell, where the initial hydraulic head remains the same throughout the simulation. A fixed head cell can be applied wherever an aquifer is in direct contact with a river, stream or lake where the water level is known to be constant. These cells can act as either an inexhaustible source or sink of groundwater. Upon simulation, the model is able to calculate whether or not the boundary is acting as a sink or a source, dependent upon initial hydraulic head values (pers. comm. Rozlapa, 2007). For this model, the seasonal river beds observed in the northeast and southwest corners, along with small creeks along the south, have been assigned as fixed head boundaries. The cells are given a value of ‘-1’ and are coloured dark blue, as shown in Figure 4.6. The northern boundary crosses highly variable topography and cuts across a major drainage structure. This has been incorporated into the model, with outflows indicated using a fixed head boundary over this drainage channel. The north-eastern boundary is some distance away from the mine area, consisting of shallowly dipping Marra Mamba Iron and Jeerinah Formations. The southern boundary follows the southern limb of the syncline, comprised of the Marra Mamba Iron and Jeerinah Formations. The western boundary cuts through the foot of the syncline and thus through most of the modelled units. This boundary acts as an outflow for water drainage, which has been incorporated into the model as a fixed head boundary. Figure 4.7 displays a 3D view of the model, with mining areas, faults and topographic features to help visualise the model setup. 50 Chapter 4: Mt. Tom Price Groundwater Model 1 -1 1 -1 0 1 1 -1 0 1 -1 -1 0 1 1 -1 -1 0 0 0 0 0 -1 -1 0 0 Figure 4.6: IBOUND array of Mt. Tom Price model in PMWIN Pro. Insert displays cell values and their corresponding colours (1 = active (white), -1 = fixed head (blue), 0 = inactive (grey)). 4.3.1.3 Initial & Prescribed Hydraulic Head In order for a flow simulation to start, initial water table values need to be defined at the time of flow commencement. This was achieved in two ways. Firstly, the original ‘static’ water table prior to pumping was used. In order to create this data, groundwater levels from 1994 were entered into a comma-separated value (.csv) file, compiled from existing database records (Rio Tinto Hydrogeological Database), as well as records from a 1994 hydrogeological report (Preston, 1994). The data can then be gridded using the SURFER software package to create an ASCII matrix grid file (.dat or .txt) that can be directly entered in to PMWIN Pro. As most of the data points are centred in and around the pits, data points towards the edges of the models extent were estimated from regional hydraulic head and topographical trends. This was done to ensure the generated grid was representative of measured regional water level values. 51 Chapter 4: Mt. Tom Price Groundwater Model ellow lines represent pit ograph of the mine superimposed. Dotted y mine superimposed. of the ograph lustrative purposes, they have been made have been made they il lustrative purposes, are inactive (grey). For of the model boarders the that note to It is important Red lines representlocations. major faults. the model setup. visualize help to active Price Groundwater Model with an aerial phot Price Groundwater Tom Mt. the of view 3D Oblique 4.7: Figure 52 Chapter 4: Mt. Tom Price Groundwater Model Due to size inconsistencies with PMWIN Pro and the exported surfer grid, some values could be assigned to zero in the initial water level matrix. These values could create problems as they would dry up immediately as flow commences. Therefore, these cells were manually entered directly into the matrix editor in PMWIN Pro. The second approach was to use steady-state model outputs as the initial hydraulic head for transient model simulations (pers. comm. C. Moore, 2008). This was achieved by running the completed model over an infinite period of time so that a state is reached where hydraulic head variations are as close to zero as possible. Appendix 5 displays the results in comparison with surfer generated values. The main difference between the two is the geometry of the perched groundwater mound, which cannot be represented accurately in this model. Due to the tight nature of the regional hydrogeology and the lack of influence on initial results, the mounded water values have remained in the initial modelled heads file and should have a minimal impact on results. Hydraulic heads to the south of the mine area are increased in this second approach, possibly indicating incorrect parameterisation or model conceptualisation, and will impact to some degree on the reliability of the estimated parameter estimates and predictive model outputs. 4.3.1.4 Grid Design In order to achieve the required detail during a simulation, the grid is set up to include smaller sized cells around areas of interest’. To avoid numerical errors during simulation, every redefined cell is at least 1.5 times the size of the adjacent cell, i.e. if 50 m cell sizes are used, the cell progression must be as follows (Figure 4.8) (pers. comm., K. Rozlapa, 2007): 50 x 50 34 25 34 50 x 50 x x x 50 50 50 50 x 34 50 x 25 25 x 25 50 x 34 50 x 50 Figure 4.8: Grid Spacing required for stable simulation using MODLFOW simulations 53 Chapter 4: Mt. Tom Price Groundwater Model A grid spacing of 25 x 25 m was used in and around the pit voids to represent cells that will be below the water level upon mine closure. These areas include the South East Prongs, North Deposit, parts of Centre and West Pits, Section Six and Marra Mamba West (Figure 4.9). These areas were defined by placing the initial water level surface into the geological model, and comparing them with the final pit designs. NTD WEST SEP SSIX MMW Figure 4.9: DXF image displaying final pit designs (pink) vs. rendered initial water table levels (blue). Blue areas within the pink pits represent potential final void water levels; therefore increase cell resolution is required in these areas. Initially, 6 layers were been created using geological templates at 50 m vertically spaced intervals. Due to the required detail of the pit voids and thinness of some of the hydrogeological units in the mining area, a 25 m vertical resolution was used in the MTPGM. In this case, each of the initial 6 layers was further refined to 2 layers, resulting in a model with 12 layers at 25 m thickness. At this resolution, major water bearing units should have an acceptable degree of connectivity, although thinner water bearing units may be inaccurately represented. It is recommended that layer numbers be kept as low as possible, as file size and computation time increases exponentially according to cell size numbers. Before grid refinement the regional MTPGM measured 264 x 218 cells with 6 layers. After refinement cell numbers increased to 395 x 282 cells with 12 layers. A drawback of this 54 Chapter 4: Mt. Tom Price Groundwater Model increased detail is the vastly increased computation time of approximately 4 hours for transient simulations. It should be noted that, for a model to be practical, the computation time should be no longer than 10 minutes (Hill, 2007). This enables accurate model calibration, as many model runs can be undertaken each day, rather than only two runs per day as with the MTPGM. This is ideal, but often regional models require longer computation times to minimise errors, especially in mining situations where large water balance errors can occur due to the extraction of large volumes of water from the model. 4.3.1.5 Time Discretisation A groundwater flow model must be assigned to a single or multiple stress periods for a simulation to occur. These stress periods are predefined measurements of time, measured in seconds, minutes, hours, days or years. A number of time steps can be used in each stress period. Using multiple time steps within a stress period lowers model errors, particularly when stress period durations are long. The MTPGM has 48 stress periods with 12 time steps that were specified in order to capture the dewatering of the pit voids from groundwater pumping commencement in 1994. Due to incomplete records and lack of data from 1994 to 2004, yearly stress periods (365 days) were specified using yearly averages in order to represent pit dewatering to some degree of accuracy. January to June 2004 was also averaged and represented as a single six month stress period (182 days). In order to observe detailed groundwater response, monthly stress periods with 10 time steps were defined between July 2004 - July 2007. This level of detail allows for easy calibration to see how accurately the transient model compares to real world records relative to varied pumping throughout the year. 4.3.1.6 Model Parameters The next step in creating a groundwater model is to input horizontal and vertical conductivity, storage and specific yield values to the model grid. These values vary according to geological units, faulted/folded zones, and weathered areas. Using the geological 55 Chapter 4: Mt. Tom Price Groundwater Model templates created from the geological model, hydrogeological parameters can be assigned quickly using the data editor. The template .dxf files can be viewed in the data editor window by using the ‘maps’ option. The horizontal hydraulic conductivity (KH), vertical hydraulic conductivity (KV), storage (S) and specific yield (Sy) values initially assigned to the model are shown in Table 4.4. However, it should be noted that due to the highly variable nature of the mining area, values deviate from the average values according to dip angles of the stratigraphy and the presence of faulted zones. Appendix 6 contains complete model layer illustrations displaying each modelled unit according to KH and KY values, which will have identical parameter boundaries with differing values. Storage and specific storage layer illustrations are included in Appendix 7-8. As these values are much less variable, less attention to detail was needed, requiring just 6 differing parameter values across the 12 layers. This illustrates the concept that groundwater models are based on hydrogeological parameter differences, rather than lithological variations. These initial parameter estimates have been altered in the calibration phase (as discussed in section 4.5), in response to the information content of the piezometric level data. Table 4.4: Model parameters used in Mount Tom Price groundwater model (incorporates modified parameter values from the Rio Tinto Hydrogeological Database (Appendix 3)). Horizontal Vertical Storage Specific Yield Parameter Unit Conductivity K Conductivity K H V Coefficient S S (m/day) (m/day) y 1 Joffre Member 0.001 0.0001 0.0002 0.001 2 Whaleback Shale Member 0.008 0.0008 0.0002 0.01 3 Dales Gorge Member 0.005 0.0005 0.00001 0.001 4 Mt. McRae Shale 0.01 0.001 0.0002 0.001 5 Mt. Sylvia Formation 2 1.7 0.000018 0.001 6 Wittenoom Formation 0.1 0.01 0.0002 0.001 Marra Mamba Iron 7 0.0001 0.0001 0.00001 0.001 Formation 8 Jeerinah Formation 0.03 0.003 0.0002 0.03 9 Fault zone 1 1 0.0002 0.001 10 Pit void 100 100 0.0001 0.99 56 Chapter 4: Mt. Tom Price Groundwater Model 4.4 MODFLOW PACKAGES MODFLOW 2000 boasts a number of packages that can be used in flow simulations. These packages are programs that communicate to the model, so that accurate representations of real world elements can be simulated. These include simple packages that apply stresses such as rainfall and pumping, as well as packages that control communication between groundwater movements through modelled cells. 4.4.1 Block Centred Flow (BCF) Package The BCF package computes conditions that determine movement rates to and from storage using Darcy’s Law. This package contains variables entered by the user that define hydraulic conductivity, storage coefficients, specific yields, leakance and aquifer cell geometry, of which each variable requires a separate array to specify details for every active cell in the grid (McDonald and Harbaugh, 1988; Facer, 1998). The BCF package allows for simulations of which a specified water table may rise into unsaturated layers. This is applicable to the MTPGM as the cessation of pumping, post mining, will result in the reactivation of previously dry cells upon hydraulic head recovery. 4.4.2 Flow Packages 4.4.2.1 Recharge In order to simulate seasonal rainfall in the Mt. Tom Price catchment boundary, the recharge package was used. This package simulates distributed recharge over the model by assigning data to the highest active cell of each vertical column of cells. Measured recharge in kL/day should be entered for each stress period, a specified time interval in minutes, hours, days or years by which the model runs. The MTPGM represented evapotranspiration with this package by entering 1% of rainfall as recharge, simulating the high evaporation rates and low table recharge in the region. The recharge rate can be defined as QR and is applied to a single cell within a vertical column of cells. The recharge flux is the physical amount of water applied to the model, defined as IR. QR = IR · DELR · DELC Where DELR · DELC = map area of a model cell (Chiang, 2005). 57 Chapter 4: Mt. Tom Price Groundwater Model 4.4.2.2 Wetting Capability The MTPGM simulates the effects of pumping on the regional water table. In order to simulate the rising water table response upon mine closure, a rewetting package must be used to lower simulation errors (pers. comm., K. Rozlapa, 2007). This is achieved using the wetting capability of the Block Centred Flow 2 (BCF2) package, allowing for the water table to rise into unsaturated (dry) model layers. When the water level falls below the bottom elevation of a cell it becomes dry and essentially inactive. When a cell becomes dry, IBOUND is set to 0, indicating no flow within the cell, and thus the conductivity values are set to 0. As a result, water is unable to flow into the cell during the simulation and the cell remains inactive, despite rising water levels (pers. comm. Rozlapa, 2007). A solution to this problem has been incorporated into the BCF2 package so that inactive cells can be turned back into active cells to allow for the completion of a flow simulation. MODFLOW contains code known as wetting threshold (THRESH), used to decide whether a dry cell can be turned into a wet cell, as follows: • If THRESH = 0, the dry cell or the inactive cell cannot be wetted. • If THRESH < 0, only the cell below the dry cell (or inactive cell) can cause the cell to become wet. • If THRESH > 0, the cell below the dry cell (or inactive cell) and the four horizontally adjacent cells can cause the cell to become wet. When a cell is wetted, its IBOUND value is set to 1 (variable head) and the hydraulic head (h) is calculated using one of the following equations, which can be adjusted in the wetting capability dialog box: h = BOT + WETFCT · (hn − BOT) h = BOT + WETFCT · |THRESH| where hn is the head at the neighbouring cell that causes the dry cell to wet and WETFCT is a user-specified constant called the wetting factor. BOT is bottom of the cell (Chiang, 2005). 58 Chapter 4: Mt. Tom Price Groundwater Model In the case of the MTPGM, the wetting threshold has been set to –5, with a wetting factor of 0.1 (pers. comm., K. Rozlapa, 2007). This was selected as there will be many cells that will be rewetted upon mine closure, when pumps are turned off and cells are rewetted from below during water table recovery. 4.4.3Well Package To simulate an injection or pumping well in a groundwater model, individual cells or polygons can be specified as sources or sinks during a specified stress period. MODFLOW assumes that the well penetrates the entire thickness of the specified pumping cell, so rates must be adjusted according to pumping levels if multi layer pumping exists (Chiang, 2005). The MTPGM includes a total of 26 pumped wells that have been active at various stages of the 14 years of pumping to date in the mine area. Due to inaccessibility of old pumping locations and volumes, some historic wells have been left out of the groundwater model. Pumping rates were averaged from existing pumping plans and flow meter readings to obtain kL/day values for each pumping well in a defined stress period. Due to availability of historic pumping records, pre-2004 pumping rates were estimated from dewatering plans, reports and flowmeter readings. Due to a lack of historic pumping records, a total of 17 pumped wells were ultimately used in the groundwater model pumping from ~10–3600 kL/day. Missing pump records were estimated and incorporated into existing pumping rates. Pumping records are presented in Appendix 9. Pumped water from the South East Prongs pit has been discharged into the Section Six pit void since 2005. This has been incorporated into the groundwater model as a single, positive pumping well, with average values of ~1000 kL/day discharged into the SSIX pit void (layer 8). 4.4.4 Solvers In order to calculate hydraulic head within a finite difference grid as used in the MTPGM, MODFLOW prepares a finite difference equation for each cell. This equation expresses the relationship between the centre of the cell and all six other nodes in the adjacent cells. As hydraulic heads can vary from one iteration to the next, the equations across the entire grid must be calculated simultaneously for each time step (Chiang, 2005). 59 Chapter 4: Mt. Tom Price Groundwater Model 4.4.4.1 Preconditioned Conjugate Gradient 2 (PCG2) Package The MTPGM has 48 time steps, of which a number of outer and inner iterations need to run in order to calculate a final value within a specified closure criterion. In order to keep modelling times down, a minimal number of iterations need to be calculated so that calculations are fast, while maintaining an accurate outcome. Due to high pumping rates during stress periods where little to no recharge has been applied, large water balance errors occur using a low number of iterations. To overcome this, 100 outer iterations and 50 inner iterations have been used. A tight closure criterion has been applied to the model to further increase accuracy, with a Head Change value of 0.01 L, and a Residual of 1 kL/day. This closure criterion represents an adequate/high error in MODFLOW computations. Ideally these values should be one or two magnitudes lower than acceptable field measurement error (Hill, 2007). Due to the nature of this model, the closure values used were the lowest possible that produced convergence. This was likely due to high conductivity and storage contrasts, as well as the large number of dry cells in the model. 4.4.5 Output Control The primary output of a MODFLOW groundwater simulation is the run listing file output.dat, which displays a volumetric water budget calculation for the entire model at each time step, providing indications on the continuity of the model. The difference between total inflow and total outflow should equal the total change in storage, displayed as percentage discrepancy values to help identify large errors during the simulation (Chiang, 2005). Percentage discrepancies higher than ~1% should be avoided and variables within the stress period should be reviewed as necessary to eliminate these discrepancies (pers. comm. Rozlapa, 2008). PMWIN Pro allows for the adjustment of certain output data via the output control including output terms, frequency and predefined head values. Certain versions of Modflow do not support budget.dat output files of greater than 2GB in file size without running into a number of errors. This file contains code for detailed inflow and outflow of the model at each time step, therefore files can become extremely large. To overcome this, the output control needs to be modified so that results from say, the last time step in each stress period is recorded, rather than every time step, minimising file size (pers. comm. Rozlapa, 2008). 60 Chapter 4: Mt. Tom Price Groundwater Model 4.4.6 Head Observations In order to include observed hydraulic heads at specified observation piezometers, the head observation dialog box can be used. Here, easting and northing coordinates can be entered along with observed hydraulic heads at a specified stress period. PMWIN Pro supports multiple-layer observations, specified with the use of the Layer Proportions table. Here, the screened interval of the piezometer can be defined by assigning the specified layer a non-zero proportion value. A full list of observation bores has been included in Appendix 10. 4.5 MODEL CALIBRATION 4.5.1 Introduction Calibration is the process of determining optimal hydrogeological parameters so that the results most closely match field measurements of hydraulic heads and flows. The purpose of calibration is to ultimately make more reliable predictions with the model than would have been possible without undertaking the calibration process. When considering a model that most accurately represents a groundwater system, an acceptable factor of error needs to be taken into consideration (Kresic, 1997). In the case of the MTPGM, a semi-regional aquifer system, a ‘tolerable’ head difference could be assumed as measuring less than 1 m. The quality and quantity of reliable field data is highly influential when calibrating a groundwater model. Due to the mine planning procedures in place at Mt. Tom Price, including weekly hydraulic head measurements and pump data collection, an excellent series of piezometric data are available. The hydrogeological parameters obtained from the Rio Tinto Drillhole Database and interpretations of geological data provides guidance, though variations from these values can be expected in between the sampling locations, as is typical in any moderately heterogeneous geological strata. The purpose of this groundwater model is to understand regional groundwater flow through the mining area and to calculate the final groundwater levels post-pumping. This can be achieved to some extent with the use of steady-state calibration, comparing hydraulic heads against pre-pumping levels. To further ensure accuracy, current water levels can be calibrated with pumping, which can easily be entered into the same model used for post-pumping simulations. 61 Chapter 4: Mt. Tom Price Groundwater Model Steady state prediction of the final pit lake levels provides good elevation information, but does not allow rates to be explored. To overcome this, a transient simulation using measured pumps was used in this project as discussed in section 4.3.1. This is helpful in improving accuracy, and also allows calculation of the rate at which the water table recovers after pumping. 4.5.2 Results A comparison of model outputs and measured water levels (hydrographs) can be found in Appendix 11. Initial results suggest that the model is somewhat accurate in certain areas, showing similar responses (both in timing and magnitude) to stresses over the specified simulation period. However, in most areas, simulated water head levels are offset from the measured levels by about 10-20 m. Observation levels in the Marra Mamba, West Pit and some locations in the South East Prongs simulate stress response most accurately. A moderately calibrated observation bore (NW5) from the north wall of the SEP has been included as Figure 4.10. The curves display similar responses the stress, but are at a constant 3-4 mRL difference. Areas of intense pumping, such as the North Deposit and the south wall of the South East Prongs, did not produce accurate results. An example of a poorly calibrated observation bore (PZSEP03) on the south wall of the SEP has been included as Figure 4.11, showing very little variance to high pumping rates. Observation levels pre-2004 are also poorly represented across the area, likely due to poor pumping and available observation records during this period. Initial results also indicate that there may be some error in initial head levels, seen in large differences in starting values in most of the hydrographs (water head vs. time curves). This suggests error in initial head surface construction, likely due to low coverage of piezometers in certain areas in the mine. Observation levels taken on and around the perched water body near the Southern Ridge area were exported for further analysis. As expected, simulated levels were generally around 700 mRL, much lower than the measured values of up to 820 mRL, and also displayed little response to stresses. 62 Chapter 4: Mt. Tom Price Groundwater Model Figure 4.10: Hydrograph example of a moderately calibrated observation hole. The blue line represents the calculated modelled output. The red line represents real measured values. NB relatively even mRL separation (~3-4 m) with downward trend. Refer to Appendix 10 for bore details. Figure 4.11: Hydrograph example of a poorly calibrated observation hole, displaying large errors (up to 30 m) with a poor response to stresses. The blue line represents the calculated modelled output. The red line represents real measured values. Refer to Appendix 10 for bore details. 4.5.3 Calibration Considerations When calibrating to observation holes near pumping wells it is important to take the inaccuracy of water levels close to or adjacent to pumping cells into the consideration. This is due to the large size of the head gradient near a well node, which itself is not modelled 63 Chapter 4: Mt. Tom Price Groundwater Model accurately as the model extracts or injects water to the entire cell rather than the nodal point. Therefore, calibration holes should be sampled from nodes away from the point source or sink (Anderson & Woessner, 1991). The groundwater mound in the centre of the modelled area is likely to cause problems upon calibration of observation holes found in the Southern Ridge and Centre Pit areas. The mound is consensually thought to be a perched water body that seems not to be hydraulically connected to groundwater source of the large scale pumping occurring in the SEP and NTD. MODFLOW simulations assume saturated hydrogeological conditions. As a result, model simulations will not represent this mound accurately due to the presence of saturated material below the water body. Therefore, calibration holes situated on or near this groundwater mound will not reflect real world observation levels and were not included in the calibration process. Pre-2004 piezometer networks in the Mt. Tom Price area were comprised of a combination of open standpipe, gravel packed and screened piezometers. Open standpipe and gravel packed piezometers are not likely to reflect model simulation results due to the fact that there may be pressurized and perched water contamination (pers. comm., L. Campbell, 2008). Since 2004, steel cased piezometers have been predominately used for increased accuracy of aquifer properties. Due to the unavailability of pumping records pre-2004, pumping rates were estimated from various pumping reports and plans. As a result, large pumping variances and changes to pumping routines may have not been accurately entered, resulting in unreliable simulation results in certain areas of the model during this period. 4.5.4 Trial and Error Calibration Trial and error calibration involves the adjustment of various hydrogeological parameters such as conductivity, storativity and specific yield values to fit the calculated results to the measured field observations. This can end up being a very long process, and can produce results that are fairly subjective. 64 Chapter 4: Mt. Tom Price Groundwater Model The MTPGM has a large number of cells and therefore a long simulation run time to produce reliable results (~4 hours). As a result of this long simulation time, extensive trial and error calibration is impractical as the time required to produce good results is prohibitively long. 4.5.5 Parameter Estimation (PEST) An alternative to trial and error calibration is the use of parameter estimation software such as PEST (v.4, Doherty, 2004). PEST assists in data interpretation, model calibration and predictive analysis by adjusting model parameters so that the fit between model outputs and field observations is optimised. This is achieved through multiple model runs and calculations of mismatches between the model output and measured values (Doherty, 2004). PEST can be run directly in PMWIN Pro when the correct parameters are entered into the model during parameter input. Alternatively, PEST can be run outside of PMWIN, using the specified parameter values as assigned to the MODFLOW cell array. This method is often an advantage and allows the most advanced PEST functions to be used. To set up PEST for use outside of PMWIN, parameter arrays must be exported to an ASCII grid file from inside PMWIN. In the case of the MTPGM, horizontal and vertical hydraulic conductivities, storage coefficients and specific yield values were exported using the matrix editor. These must be exported from their individual layers and named accordingly, e.g. hk1 (horizontal conductivity for layer 1), vk8 (vertical conductivity for layer 8), sc4 (storage coefficient for layer 4), sy12 (specific yield for layer 12). Observation heads and well location files should also be created in the correct .txt format so that PEST can function correctly. As the generated geological model of the area is highly detailed, each parameter unit is defined as an individual parameter. PEST calculates values for each individual parameter. PEST uses information from the ASCII grids, piezometer readings as well as the various MODFLOW packages within the MTPGM exported from PMWIN. Using the ASCII grid files, command files need to be created so that PEST can interact with them. Parameter value (.INF) files need to be created to identify the different parameters in each layer according to the MODFLOW grid matrix. In total, there are a maximum of 22 65 Chapter 4: Mt. Tom Price Groundwater Model numerical parameters or hydraulic conductivity, and 6 parameters for storage and specific yield. Therefore are a total of 45 different hydrogeological parameters within the entire model. These parameters need to be listed and referenced in template files, so that PEST can assign each parameter to its correct value, e.g. hk1 = 1, vk2 = 2, sc3 = 0.0001, sy4 = 0.01, etc. Once the template and instruction files have been prepared, a PEST control file needs to be created which “brings it all together” and “tunes” PEST to the case in hand (Doherty, 2000). As there are 45 model parameters in total in the MTPGM, PEST must run the model at least 45 times for the initial iteration. The 45 parameters can then be grouped into ‘super parameters’ in order to minimise the large number of model runs required. From initial results, we can inspect the degree if parameter dependency by inspecting the eigen components of the model normal matrix. This indicates around 10 super parameters will be sufficient to represent most parameter detail (Figure 4.12). The calibration problem was then redefined in terms of the 10 largest eigen components of the model normal matrix, via the SVD-Assist regularisation utilities of PEST (pers. comm.., C. Moore, 2008). The PEST simulation was continued by creating a second super parameter control file, and then running the PEST optimisation as before. Figure 4.12: Super Parameter Selection 66 Chapter 4: Mt. Tom Price Groundwater Model Initial PEST outputs suggest that hydrograph variances had tightened up to some extent. Responses to variable stresses remained similar to initial, non-calibrated results although the some large starting head variances were still apparent. Results from the initial PEST runs have been included as Appendix 12. Here the optimisation iterations are listed with the optimised horizontal, vertical hydraulic conductivity, storage and yield values presented. One of the most notable changes during optimisation is the Mt. Sylvia formation (hK1), which has been lowered from a KH value of 2 m/day, to 0.75 m/day. Pit void values were also changed dramatically upon PEST optimisation. Values changed from 100 m/day to 68 m/day, which is acceptable as this parameter has essentially been given an arbitrary high value to represent open air in the pit void. Storage and yield values were also altered, but still within an acceptable representation of open air. Results from the super parameter estimation have been included as Appendix 13. These results show slighter less variation to the initial parameters optimisation runs, with few notable differences such as hk2, where the values have dropped from 0.75 m/day to 0.13 m/day. This parameter represents the Mt. Sylvia formation and can be considered reasonable as values of 2 m/day should only be represented around intensely folded and faulted zones. 4.5.6 Final Calibration To overcome the large head variances seen in the initial hydrographs, initial head values were obtained by running a steady state model that simulates pre pumping conditions. This helped to lower the gaps between the hydrograph curves for the best fit possible. There was however several areas where the steady state head outputs either far too high and low according to the measured heads. To overcome this, initial heads were manually adjusted in PMWIN to match initial measured heads as seen in the measured heads records. It became apparent that certain areas such as the South East Prongs, North Deposit and Section Six were not responding well to stresses. It is essential that these outputs match the measured values as close as possible so that an accurate calculation of final pits values can be determined. It was decided that a combination of PEST optimisation parameter input and trial and error calibration would produce the best results with limited time available. 67 Chapter 4: Mt. Tom Price Groundwater Model Using the template files (.inf) created for PEST calibration, super parameter outputs were entered into the grid matrix by simply replacing the parameters listed in the output file. These layer files were loaded back into PMWIN via the matrix editor. From here, areas of concern were adjusted to reflect real world stress response. This usually involved the adjustment of horizontal, vertical hydraulic conductivity and storage values. As PEST approximates overall values for an individual parameter, which is this case is generally the entire formation; zones of faulting and folding were most likely to have been overlooked. Using the matrix editor, strongly folded and faulted formations were given horizontal and vertical conductivity values of 5-7 m/day. Several historical pumps and piezometers used in this model were screened within the mineralised ore. Therefore it was necessary to represent the ore accordingly. To produce optimised calibration in both the old and recent localities, pit floor levels of mid 2006 were used, with ore given horizontal and vertical conductivity values of ~10 m/day. Higher storage and specific yield values were also applied to help represent the rapid drawdowns seen in response to intense pumping regimes. These values were approximated from database records, as well as from Preston’s Tom Price Hydrogeological studies (Preston, 2004), of which, detailed pump tests of the various historical water bores have been presented. Results from the final calibration display a great improvement, representing the dewatering of the pits sufficiently. A full record of hydrographs has been included in Appendix 14. The results suggest that the hydrogeology of Marra Mamba West Pit has been the accurately understood and entered into the model. Responses here are excellent, deviating from the real measured values by less than 1 m towards the end of the simulation at observation bore PZ5MM2 (Figure 4.13). Hydrographs collected from the South East Prongs area displayed a huge improvement. Dewatering was accurately represented at observation bore PZ03SEP3, following the downtrend with maximum deviations of approximately 5 m (Figure 4.14). There are however several areas where hydrogeology was not sufficiently represented and resulted in several hydrographs that did not reflect real measurements. Several bores sampled 68 Chapter 4: Mt. Tom Price Groundwater Model from the North Deposit dewatered far too rapidly, resulting in deviations of the calculated values by up to 35 m (Figure 4.15). Figure 4.13: Hydrograph example of a well calibrated observation hole. Simulated head levels at PZ5M2 respond precisely to measured levels with only 1-2 m offset. Refer to Appendix 10 for bore details. Figure 4.14: Hydrograph example of a moderately calibrated observation hole. Simulated values at PZ03SEP3 follow the general downward dewatering trend, with maximum offsets of approximately 5 m. Refer to Appendix 10 for bore details. 69 Chapter 4: Mt. Tom Price Groundwater Model Figure 4.15: Hydrograph example of a poorly calibrated observation hole. Simulated values at PZ6NTD7 are offset by approximately 35 m, showing a different stress response to the available measured data. Refer to Appendix 10 for bore details. 4.6 SENSITIVITY ANALYSIS &MODEL PREDICTION UNCERTAINTY It is often useful to assess the reliability of a groundwater model by analysing the sensitivity of the various parameters used in the groundwater model. PEST optimisation model runs calculate these sensitivities and allow for quick visualisation of parameters that are particularly sensitive to the slight parameter adjustments PEST applies to the model. PEST runs of the MTPGM approximated that a horizontal hydraulic conductivity parameter hk11 (0.05 m/day), an arbitrary value assigned to various units to lower parameter contrasts across a model layer, was the most sensitive to PEST optimisation. This is understandable as this value was hydrogeologically unrepresentative but was used to stabilise the model. Low sensitivities were observed in typically tight units such as the unmineralised Dales Gorge Member (hk9 at 0.03 m/day). Observation location sensitivities can also be analysed in order to define areas of a model that respond well to parameter adjustments. The Marra Mamba West observation bores were the most sensitive, indicating this area has been modelled to a good degree of accuracy. Observation bores in around the South East Prongs pit display low sensitivity values, reflecting the inaccurate and flat water table response as seen in the hydrographs at these locations (Appendix 14). PEST parameter sensitivity outputs have been included in Appendix 15. 70 Chapter 4: Mt. Tom Price Groundwater Model The final calibrated MTPGM uncertainty can be assessed by visualising the final calculated outputs against measured values. This is a similar method to calibration, but allows for the quantification of the accuracy of the simulated outputs. Ideal results should be plotted as a linear trend line with gradient and correlation values as close to 1 as possible. Results indicate a relatively low correlation value (R2) of 0.337 (Figure 4.16). This value has been lowered significantly due to errors associated with the misrepresentation of dewatering of the NTD and SEP at several observation localities, as discussed in section 4.7.8. The calibrated model and PEST files have been included as Appendix 24. Figure 4.16: MTPGM uncertainty displaying measured heads against modelled heads. 4.7 PREDICTIONS 4.7.1 Prediction Model Setup The primary aim of this thesis is to assess groundwater conditions in the area upon mine closure. Using the calibrated MTPGM, a prediction on pit lake recovery can be made by simply creating a new prediction model MTPPGM using the same parameters. To represent final mine conditions, final pit designs were entered into the model, using the same pit void parameters as previously used. This model was assigned 40 new stress periods to observe final pit recovery up to 200 years after the cessation of mine dewatering. These periods as assigned as 1 year durations, progressing to 5 year durations, and finally 10 year durations. 71 Chapter 4: Mt. Tom Price Groundwater Model To simulate recharge over the recovery period, average precipitation values of 300 mm/year was assigned to the model. Evapotranspiration and depth to water table was accounted for using only 1% of this value as a direct input to the modelled groundwater water table. To produce realistic initial heads, outputs from the 48th stress period (June 2007) of the transient MTPGM were imported as initial heads in the new MTPPGM. To represent final pit conditions of the Marra Mamba West Pit, a cone of depression was entered into the initial heads values, reaching 15-20 m below the final pit design of this mine. A primary goal of this thesis is to calculate recovery and final pit lake levels upon mine closure. Therefore, final water levels according to final mine designs should be simulated. This will produce recovery curves that can be used to determine final pit lake levels and recovery rates. Using the modified June 2007 hydraulic head output file, final water levels were entered as cones of depression, approximately 15-20 m below all final pit designs. A second model was also set up with levels representative of final pumped conditions upon mine closure. Cones of depression have been extended 20 m below all final pit designs (excluding Section Six, which is already recovering) to represent accurate pit closure recovery (Appendix 16). 4.7.2 Results In order to approximate the best results from the modelled outputs, it is essential that all data is assessed and taken into consideration. From the transient results of the calibrated model, it was clear to see that some areas of the model do not fully reflect real world conditions. This is due a combination of misinterpretation of parameters and/or resolution capture of essential hydrogeological units. As most of the hydrographs respond sufficiently to stresses, it is usually necessary to simply adjust mRL values so that the responses are as close to real world results as possible. This is normally done through further calibration, but in the case of this thesis, after long attempts at calibration, mRL adjustment was deemed practical. Final heads can be visualised in Appendix 16. Recovery curves from the June 2007 levels were initially successful in early periods of the simulation. However, it is apparent that final recovery levels as about twice as long as estimated from rule of thumb recovery times (~100 years). The results have been summarised as Table 4.5 and included in Appendix 25. Results suggest that the North Deposit will be the fastest to fully recover, reaching levels of 730 mRL in 140 years. Marra Mamba West and 72 Chapter 4: Mt. Tom Price Groundwater Model West Pit are the slowest to recover, refilling to levels of 665 mRL and 715 mRL in 180 years. Recovery curves from final pit pumping values have been included as Table 4.6. Final pit floor designs have been added to results to visualise pit lake depths with time. Interestingly, despite lower pumped levels, full recovery occurs faster than the June 2007 scenario. Section Six is the fastest pit to fully recover, reaching levels of 662 mRL in 92 years. The North Deposit is the slowest to recover, refilling to 730 mRL in 120 years. Table 4.5: Predicted long term pit lake conditions at Mt. Tom Price from June 2007 recovery scenario Pit Predicted Long Term Conditions Predicted Pit Time Taken for Time Taken For Void Water 50% Recovery Full Recovery Level (mRL) (Years) (Years) 730 22 140 North Deposit 660 20 150 South East Prongs 665 23 180 Marra Mamba West 655 24 178 Section Six 715 75 180 West Pits Table 4.6: Predicted long term pit lake conditions at Mt. Tom Price from Final Pit recovery scenario Pit Predicted Long Term Conditions Predicted Pit Time Taken for Time Taken For Void Water 50% Recovery Full Recovery Level (mRL) (Years) (Years) 730 22 120 North Deposit 710 30 115 South East Prongs 655 22 96 Marra Mamba West 662 55 93 Section Six 700 55 98 West Pits North Deposit Pit The North Deposit recovery curve shows a gradual rise at the start of the simulation (Figure 4.17). Recovery rates increase after 14 years, likely due to permeability increases as the water table rises though the tight basement units. This increase slows down after approximately 30 years from the start of the simulation, finally reaching 100% recovery to 730 mRL after 120 years. 73 Chapter 4: Mt. Tom Price Groundwater Model South East Prongs Pit Recovery curves from within the South East Prongs pit are similar to the North Deposit, with slow initial recovery rates in the first 25 years of the simulation, increasing as mRL levels rise (Figure 4.18). The increase gradually slows down to a final level of 710 mRL after 115 years. There seems to be an error in the modelled outputs, displaying a much higher mRL value then specified. This seems to have minimal effects on results, so it can be ignored. Marra Mamba West Pit Marra Mamba West Pit recovery curves display a rapid recovery in the first 4 years of the simulation (Figure 4.19). After this, there is a gradual increase until full recovery is reached at 655 mRL after 96 years. Once again a starting head error has occurred, but can be ignored. There also appears to be significant steps in the recovery curve that coincide with modelled layer mRL values. These are likely to be caused by errors associated with the reactivation of the dry cells above rising water table. These errors appear to have a small impact on final results by raising the final pit levels by approximately 10-15 m. Section Six Pit Recovery curves in the Section Six pit display a lowering in mRL values for the first 18 years of the simulation (Figure 4.20). This is likely due to the movement of the discharged water that has been deposited into the pit over the last 3 years. This response appears to be amplified, but gives an indication of groundwater response upon pit closure. Groundwater head values stabilise and then recover to a final head of 662 mRL after 96 years. Layer rewetting issues have occurred at this locality also, raising levels by approximately 5-10 m. It should also be noted that 50% recovery occurs after 55 years, considerably longer than other localities. West Pit West Pit recovery curves appear to have large errors that are likely as a result of rewetting errors (Figure 4.21). It is however useful to observe recovery trends to give an idea of full recovery times. Recovery curves in the West Pit suggest a much more gradual increase as compared to the other pits. Full recovery to 698 mRL occurs after 98 years. The West Pit also takes a considerable amount of time to achieve 50% recovery, occurring after 55 years. 74 Chapter 4: Mt. Tom Price Groundwater Model Figure 4.17: North Deposit Pit recovery from prediction model. Green lines represent 50% and 100% hydraulic head recovery. Figure 4.18: South East Prongs Pit recovery from prediction model. Green lines represent 50% and 100% hydraulic head recovery. Figure 4.19: Marra Mamba Pit recovery from prediction model. Green lines represent 50% and 100% hydraulic head recovery. 75 Chapter 4: Mt. Tom Price Groundwater Model Figure 4.20: Section Six Pit recovery from prediction model. Green lines represent 50% and 100% hydraulic head recovery. Figure 4.21: West Pits recovery from prediction model. Green lines represent 50% and 100% hydraulic head recovery. Final Outcomes To produce realistic final results, all factors of error were negated from the final pit model simulation. These include the head difference offsets calculated from transient modelling results (Table 4.7), as well as rewetting issue associated with the recovery simulation. This correction matched earliest known observation levels in several localities. This correction however was unrealistic in the North Deposit and South East Prongs, producing mRL values much higher than initial pre-pumping values. As a result, a comprimisation between measured and calculated values was achieved to produce the most accurate results possible. From the corrections applied to the recovery hydrograph results, predicted long term conditions have been presented in Table 4.8. Using this data, Pit lake volumes can be calculated from final pit designs, which is essential in calculation of final water quality of the pit voids. 76 Chapter 4: Mt. Tom Price Groundwater Model Table 4.7: Head difference correction values from transient calibration hydrographs Head Difference Pit Correction (m) 20-30 - increase North Deposit 10-20 - decrease South East Prongs 1 - decrease Marra Mamba West 20-25 - decrease Section Six 15 - increase West Pits Table 4.8: Predicted long term pit lake conditions at Mt. Tom Price from Final Pit recovery scenario (Corrected from transient results and modelled errors) Pit Predicted Long Term Conditions Predicted Pit Time Taken for Time Taken For Full Pit Lake Volume Void Water 50% Recovery Recovery (Years) (ML) Level (mRL) (Years) 710 22 120 19,690 North Deposit 685 30 115 17,030 South East Prongs 635 22 96 1,180 Marra Mamba West 640 55 93 100* Section Six 700 55 98 ? West Pits * Estimated – no final pit .dxf file available These corrected pit levels were included in the final pit design models. Figure 4.22 illustrates final pit levels at 100% recovery levels, i.e. Mt. Tom Price Mine conditions approximately 120 years into the future. Pits affected by regional groundwater flow have been included. Areas such as the Southern Ridge and Section Seven, areas affected by perched aquifers have been left out accordingly. 77 Chapter 4: Mt. Tom Price Groundwater Model Figure 4.22: Aerial view of full pit lake recovery conditions at Mt. Tom Price Mine, Circa 2130 (Pit designs from Rio Tinto Geological Database, 2008). 4.7.3 Pit Infilling Mine closure procedures often involve the refilling of pit voids with waste material from mining practices. To assess the impact this had on groundwater recovery in the pit voids, pit infill was entered into the recovery model to levels of 645 mRL. Pit infill was given high conductivity values of 20 m/day, with a specific of 0.01 and storage values of 0.0001. Recovery curves were unsuccessful, failing to represent similar conditions expected to the pit void recovery model. Results suggested that there was an issue with rewetting of dry cells as the water tabled recovered through the pit material. North Deposit recovery curves are representative of recovery trends throughout the mine (Figure 4.23). A flattening off of recovery curves occurs in the model during the period of 7-20 years after the start of simulation time. This is likely caused by sharp parameter contrast the rising water table encounters as it reaches the top of the pit infill material. A continual increase in groundwater levels occurs until approximately 130 years recovery time. This was immediately followed by an extreme drop in groundwater values (10-80 m), due to rewetting errors. A complete list of recovery curves has been included as Appendix 26. 78 Chapter 4: Mt. Tom Price Groundwater Model Figure 4.23: North Deposit recovery curve from prediction model with pit infill to 645 mRL. 4.8 MODEL LIMITATIONS Groundwater modelling attempts to recreate and simulate real world hydrogeological parameters as best as possible. Accurate interpretation of units and stresses is essential and without this knowledge, the integrity of a groundwater model becomes compromised. The MTPGM and MTPPGM have been constructed from drill-hole database records, interpreted cross sections and field measurements. All of these datasets are prone to human error upon collection, such as misinterpretations of geological units at drill depths and misrepresentation of geological structures such as faults and folds. This is likely to have been reflected in the modelled parameters, amplified due to loss of data from the creation of the geological template files from the 25 cross sections available. Groundwater model resolution is important in representing hydrogeological units accurately. In the case of the MTPGM and MTPPGM, the 25-50 m cell size was more than adequate in capturing regional hydrogeological features, but proved to be too coarse to correctly capture the complex, highly folded, thin aquifer systems in and around the pit voids. This was a trade off between accuracy and modelling times. The results suggest that representation in and around the pits voids were adequate, and further studies could represent these detailed hydrogeological connections more accurately. The lack of accessible hydrogeological information of the various structures throughout the mine has likely had an impact on the integrity of the MTPGM and MTPPGM. Folding and faulting can cause permeability augmentation of units such as the Mt. Sylvia Formation, 79 Chapter 4: Mt. Tom Price Groundwater Model causing groundwater fluctuations around these areas of concern. These areas have been as represented as best possible with the limited documented data available. Hydrogeological stresses imposed upon the model can become misrepresented due to the length of stress periods used. Looking through the pumping and rainfall records, there are periods of intense rainfall and pumping that are short in duration, followed by much lower values for the rest of the month. This usually occurs during cyclonic activity for example. As stress periods use averaged pumping and rainfall inputs, averaged over a period of 30 days, short term misrepresentation can be expected, which could be reflected in various hydrograph records. The MTPGM uses low recharge to simulate evapotranspiration and low water table recharge influence. This is a very basic representation of true evaporation and recharge rates, ultimately comprising stress responses observed during model simulations. This rate has large implications on the reliability of the final pit lake elevations as factors such as pan evaporation rates will not be represented. 4.9 SYNTHESIS Before setting up the MTPGM, it was helpful to explore previous groundwater models to give an indication on prediction times for recovery, as well as parameter consistency. The MTPGM was setup using PMWIN Pro, a graphical user interface for use with MODFLOW. Hydrogeological parameters were entered via geologic template files exported from the geological model. Stresses such as recharge and pumping were entered via software packages within MODFLOW. The model was run to simulate measured 1994-2007 responses to stresses as best possible. Output files from observed piezometers were studied for calibration purposes. A Parameter Estimation (PEST) software package was used to with the model to try and lower the stress response variances that were seen in the model output files. This was achieved by the adjustment of hydrogeological parameters such as conductivity and specific yield values. Using the calibrated MTPGM, a prediction model simulation of final pit lake recovery was created (MTPPGM). Recovery curves predicted that full recovery of the hydraulic head in the pit voids varied from 96 to 120 years, recovering to levels close to the initial heads measured in 1994 before large-scale pumping commenced. 80 Chapter 5: Hydrochemistry CHAPTER 5 MT. TOM PRICE HYDROCHEMISTRY 5.1 INTRODUCTION Water chemistry data from production and monitoring bores was studied in order to help further understand groundwater flow in the Mt. Tom Price mine. This analysis was carried out to assist the characterisation of aquifers and to determine any possible interconnection between bedrock aquifer flow systems. Water sampling is undertaken from wells by on-site hydrogeologists and environmental teams on a regular basis. Weekly pH, conductivity and flowmeter values are sampled from functional water bores, while more detailed chemistry samples are often measured every 3 months. This data is stored in the Rio Tinto EDMS Database and was exported for data manipulation. Samples were taken from locations with active sampling and evenly distributed across the entire mining area. Screened geological information was also important to classify the effects the different units had on water chemistry. Locations that had good time series data were preferred so that changes in chemical composition and pH with mining progression could be observed over time. Predicted water quality information on the final pit voids upon pit closure is important for environmental planning purposes. Using predicted pit lake levels from the MTPPGM, along with chemical data and previous reports, an estimation of final pit lake water chemistry can be determined. The Mt. McRae Shale unit lies immediately below most of the minable reserves, leaving vast quantities of the exposed unit once mining has ceased. This unit is typically synclinal in 81 Chapter 5: Hydrochemistry structure, resulting in potentially large oxidation and acid forming surfaces in the remaining pit voids. The South East Prongs, Southern Ridge and Section Six pits are areas that have significant exposures of Mt. McRae Shale, creating high levels of acidity within these pit lakes. 5.2 PREVIOUS HYDROCHEMICAL WORK A number of hydrochemistry studies have been undertaken on the Mt. Tom Price mine area. These reports are focussed mainly on long term hydrochemistry conditions of the pit lakes, which are highly influenced by the acid generating Mt. McRae Shale. A complete list of the reports has been included as Appendix 27. 5.2.1 Aquaterra Hydrochemical Modelling (Hall, 2002) In April 2002, Pilbara Iron commissioned Aquaterra to undertake a study on final pit void quality using outputs from the Aquaterra SEP and NTD groundwater models previously described in Chapter 4. Hydrochemistry data was taken from the Rio Tinto EDMS database, the same records as used in this thesis, which will be discussed in detail in section 5.3. A complete collection of all have been included as Appendix 27. It has been predicted in this study that the water quality of all the pits would tend towards brine waters (greater than 100,000 mg/L (Drevor, 1997)), with acid waters forming where large amounts of black shale are exposed (Table 5.1). The pits were determined as ‘sinks’, as most of the outflow will be a direct result of evaporation, while the solutes are retained within the pit lake (Hall 2002). Table 5.1: Long term hydrochemical predictions post mining (Hall, 2002). Predicted Downstream Pit Predicted Long Term Hydrochemical Conditions Impacts Predicted Final Pit Hydraulic Connection Salinity After Lake Other with Aquifers 100 years Salinity (mg/L) No Black Shale expected. Minor salinity increase North Partial 6,100 Brine Dominant water quality driver is downstream of pit in Deposit Sink/Throughflow evaporative concentration Hardey River Aquifer South East “Dry” NA Brine Acid salt pan predicted Negligible Prongs Sink Between Pit water expected to be acidic Southern 35,000 to regardless of fill level due to Ridge Sink? 115,000mg/L, Brine large amount of Black Shale ? dependent on exposed and limited neutralising final fill level. capacity of groundwater. 82 Chapter 5: Hydrochemistry These results were approximated using models that predict considerably lower final pit lake elevations compared to the MTPPGM outputs. Modelling undertaken by Aquaterra assumed that water table recovery is likely to be low, which is not the case with the MTPPGM, which predicts levels of 685 mRL in the SEP (compared to the previous modelled value of approximately 550 mRL), and 710 mRL in the NTD (compared to the previous modelled value of 662 mRL). Therefore the groundwater quality results of this thesis will be expected to be more ‘diluted’ than the values presented in the Aquaterra report. 5.2.2 EWL Sciences Water Quality Assessment (Jones, 2002) In conjunction with Aquaterra in February 2002, EWL Sciences undertook a future water quality study, focusing on the South East Prongs and Southern Ridge pits. This was achieved by taking a number of factors into consideration, including the pH-buffering capacity of pit materials, rainwater chemistry, the acid-neutralisation capacity of pit water and the buffering capacity of other inflows (e.g. carbonate in groundwater) (Jones, 2002). The Southern Ridge Pit lake water is sourced from a perched aquifer system and was therefore unable to be studied using the MTPPGM. Studies revealed that only the Mt. McRae Shale had strong acid water formation potential, especially when in contact with air and water. The SEP pit south wall has high acid forming potential, due to the abundant presence of high sulphur containing Mt. McRae Shale below the main ore body. Calculations on final pit quality for the studied localities have been presented as follows: Southern Ridge The final pit water quality was calculated as shown in Table 5.2. From field tests of the pit walls, an alkalinity of approximately 160 mg/L (0.16 kg/m3) was applied. Due to the surface area and amount of potential acid forming water, several pit levels were presented, showing large variations with high levels of acidity at lower water levels (Jones, 2002). Table 5.2: Final pit lake conditions according to pit lake levels for the Southern Ridge Mine (Jones, 2002) H2SO4 equiv Area of Shale Total CaCO3 100y Volume H2SO4 RL (m) Water Vol (m3) CaCO3 above pit lake acid/alkali (tonnes) of Shale (m3) (tonnes) (tonnes) (m2) 770 387066 61.9 60.7 241155 482310 86816 1430 785 1703956 273 268 197572 395144 71126 265 800 3950208 632 619 153990 307980 55436 90 83 Chapter 5: Hydrochemistry South East Prongs Calculations on final pit water quality were also undertaken for the SEP pit. A negligible <12 mg/L CaCO3 alkalinity was used in the calculations, suggesting that any inflow would have an insignificant influence on the acid conditions. These calculations also assumed that final water levels will be below the pit floor in this location, with the sole input from rainfall, producing a salt pan for most of the year. Table 5.3 illustrates the Mt. McRae Shale influence on groundwater and provides an indication of the final water quality in the SEP, should water levels reform above the final pit floor. The results suggest fairly acid conditions with high TDS values due to the high surface area of exposed Mt. McRae Shale and low throughflow rates. Table 5.3 Composition of 1:2 extract of the Mt. McRae Black Shale, Environmental Geochemistry International, 2001(reproduced in Hall, 2002)). Parameter Value pH 2.5 EC dS/m 8.12 Al mg/L 357 As 9.27 Ca 5.5 Fe 2434 Mg 16.1 Na 0.9 SO4 7368 5.3 MT. TOM PRICE HYDROCHEMISTRY 5.3.1 Background Water quality data in the Mt. Tom Price is sampled regularly by onsite hydrogeologists and environmental teams. This is essential due to the continual exposure of the acid forming Mt. McRae Shale though the extraction of the overlying iron ore-bearing Dales Gorges Member. Acidity values are sampled weekly from water bores, while more detailed elemental and TDS data are obtained monthly from water bores and piezometers. This data is available for manipulation on the Rio Tinto EDMS Database. EDMS contains approximately 200 sample locations with regular sampling records dating back to 1999. 84 Chapter 5: Hydrochemistry In mining practice, it is often useful to observe time variant water chemistry data in order to visualise any long term trends associated with continued mine activities. This is particularly useful in the Mt. Tom Price mine, as the continued exposure of the underlying Mt. McRae Shale will have a direct influence on groundwater quality with time. 5.3.2 Time Series Water Quality Data To observe any variances in long term water quality, it is helpful to plot chemical elements versus time. This will help identify any long term trends in problematic element oxidation during continued mine practice. Generally, water chemistry values seem to be consistent during the sampling data available. In some areas, values for most anions and cations drop between mid 2004 to late 2005. Since this date, there has been a steady rise in values back to the residual and in some cases a continual increase. These variances are likely due to heavy cyclonic activity in the region in this period. A complete list of sample locations and time series graphs is attached as Appendix 17. Locations with notable changes over the sampling period are summarized as follows: MM01 MM01 is an open standpipe drilled from 712–622 mRL through the Marra Mamba Iron Formation at 136672.25 E, 8186.54 N near the proposed MMW pit. Results suggest that generally, values show little variation over the period between September 1999 to August 2007. There are however notable increases in potassium and iron levels between 1999 and 2006. Potassium values reach a maximum of 98 mg/L in February 2007, from a low of 5.3 mg/L in November 2002. Iron levels fluctuate intensely since 2004, rising as high as 47 mg/L in November 2005 and falling as low as 0.02 mg/L in February 2007. The period of 2001- 2003 displays a general drop in all values, followed by a gradual increase (Figure 5.1). Section Six Pond Section Six pond is a 35 m deep pond within the abandoned Section Six pit (14000 E, 9200 N). This pond effectively screens the Dales Gorge member and Footwall Zone. Results show a gradual decrease in values between 2002-2007, with a major decrease in iron values, 85 Chapter 5: Hydrochemistry dropping from over 1000 mg/L to 0.7 mg/L during this period. Potassium levels have risen during this period, increasing from 1 mg/L to 20 mg/L (Figure 5.2). A dramatic lowering of sulfate can also be observed, dropping from 8900 mg/L to 500 mg/L over the course of 4 years. WEP05 WEP05 is a piezometer that screens the Dales Gorge and Footwall zone from 730-700 mRL in the West Pit mining area (12082.37 E, 10913.14 N). Results show minor fluctuations between 2002-mid 2007, with a relatively stable long term trend. There seems to be notable lowering of sulfate, sodium and calcium values during the period between 2004-2005 and again in mid 2006 through to mid 2007. Peak sulfate levels reached 280 mg/L in February 2004, with a low of 13 mg/L occurring in November 2006. Iron concentrations have become much more variable in recent times, especially during mid 2006 to mid 2007, where a large rise and fall of up to 0.05-3 mg/L occurred (Figure 5.3). Figure 5.1: MM01 hydrochemistry values vs. time 86 Chapter 5: Hydrochemistry Figure 5.2: Section 6 hydrochemistry values vs. time Fig 5.3: WEP05 hydrochemistry values vs. time 5.3.3 Acidity Changes Over Time Due to the acid forming nature of the Mt. McRae Shale left behind in the pit voids, it is useful to analyse pH values for any long term trends. The results should be indicative of post mining acidic conditions in the pits. Complete acidity vs. time records are included in Appendix 18. 87 Chapter 5: Hydrochemistry Generally, there has been a lowering of pH (i.e. increase in acidity) at most sample locations in the mining area, over the various documented time periods. This increase seems to have slowed down at around 2004, since which time a slightly increasing trend in acidity has occurred. Section Six can be used as a reference for the reaction of pit lake waters with exposed Mt. McRae Shale. Unfortunately there is a lack of accessible data for this area, but limited records suggest balanced pH values of ~3.5. Acidity changes with time in the mining area can be considered regionally consistent (Appendix 18). The MM01 observation location has good historical data and represents similar trends seen throughout the mining area, displaying initial lowering from high pH levels, with a gradual increase from mid 2004 (Figure 5.4). pH values are at a peak of 7.3 during August 2000, falling to a minimum of 5.82 in May 2004. Figure 5.4: Acidity sampled at MM01 vs. time with polynomial trend line added. 5.3.3 Discussion The time series results suggest that there a no major long term groundwater quality trends apparent in the data provided. Variances in cation and anion values coincide with season cyclonic activity. Evaporation, rainfall and quality of water pumped in all contribute to water quality variances as seen in the records (pers. comm., G. Domahidy, 2008). 88 Chapter 5: Hydrochemistry 5.4 CHEMICAL DATA PRESENTATION 5.4.1 Piper Plots In order to define hydrochemical facies of the natural waters in the Mt. Tom Price mine area, dominant ions were plotted on a trilinear Piper Plot diagram. Locations were selected according to operational status, with screened dates typically taken from February to August 2007. The EDMS water chemistry database from Mt. Tom Price uses milligrams per litre units, which have to be converted to milliequivalents for use with Piper Plots. Generally, there was a good spread of data points across the mine, covering most of the relevant pits and geological units (Figure 5.5-5.6). Groundwater in the Mt. Tom Price mine area can be generally classified as Magnesium type, Cl + SO4 groundwater, with a tenancy to change to Cl + SO4, HCO3 water depending on geological substrate influence. Groundwater within the Dales Gorge Member tends to exhibit lower anion concentrations, with higher levels of chloride. The Footwall Zone, however, tends to produce groundwater with low concentrations of bicarbonate and carbonate. This is likely due to the progression into the sulfate rich Mt. McRae Shale, of which the Footwall Zone is a sub-unit. Groundwater flow through the Wittenoom Formation and Marra Mamba Iron Formation characteristically has higher concentrations of Calcium with much lower sulfate concentrations. Groundwater classifications according to geological hosts as of 2008 are as follows: • Dales Gorge member: Ca + Mg, Na + K, magnesium type (cations), no dominant type (anions) • Footwall Zone: Cl + SO4, magnesium type (cations), sulfate type (anions) • Mt. McRae Shale: Cl + SO4, magnesium type (cations), sulfate type (anions) • Mt. Sylvia Formation: Cl + SO4, magnesium type (cations), sulfate type (anions) • Marra Mamba Iron Formation: Cl + SO4, magnesium type (cations), sulfate type (anions) • Wittenoom Dolomite: HCO3, Cl + SO4 89 Chapter 5: Hydrochemistry Key: + Dales Gorge member ▲ Mt McRae Shale o Marra Mamba Iron Fm. X Wittenoon Formation ♦ Mt Sylvia Fomation Fig 5.5: Piper Plot displaying hydrochemical variation according to sampled geological host – mid 2007. To delineate spatial variation across the mine site, major cations and anions were plotted according to pit sample location. Results suggest that there is little variance across the pits, with the exception of the North Deposit pit where there is clear clustering of lower cation to anion ratios on the piper plots (Figure 5.6). 90 Chapter 5: Hydrochemistry Key: + Marra Mamba Pits o South East Prongs + West Pits ■ Section Six ♦ North Deposit Fig 5.6: Piper Plot displaying hydrochemical variation according to sampled location – mid 2007 5.4.2 Stiff Plots An alternative way to define hydrochemical facies is to plot water quality data using Stiff Plots. These are formed by creating four parallel horizontal axes extending either side of a vertical zero axis. Significant cations, such as Na + K, Ca, Mg and Fe are plotted to the left of the axes, while anions such as Cl, SO4, CO3 and HCO3 are plotted to the left (Fetter, 2006). To help visualize any connectivity throughout the mining area, plots that have been sampled from similar geological units have been coloured accordingly (Figure 5.7). As some sample locations are open standpipes, or screen two separate formations, plots can show combined signatures of several screened units, resulting in contamination of the geological host rock signature. Locations were selected according to operational status, with screened dates typically taken from February to August 2007. To help visualise spatial variation in the mining area, stiff plots have been plotted according to sampled location (Figure 5.8). MM01 91 Chapter 5: Hydrochemistry and MM02 are the only open standpipes sampled, however as the Marra Mamba Iron Formation is the only unit screened at these localities, results should be considered reliable. As little information is available of screen intervals and depths, piezometers that screen multiple cannot be sufficiently determined other than through the analysis of variances in hydrochemical signatures. It is clear that geologic units have a large influence on hydrochemical facies. Clear patterns can be identified according to screened unit, with high cation levels present in the Marra Mamba Iron Formation (Figure 5.7). From the plots, the different hydrochemical facies according to screened geology has been characterised. These results should effectively produce similar results to those approximated from the piper plots. As a general rule of thumb we can define dominance in chemical elements and compounds if a particular element is higher than 50% the second highest value The Dales Gorge Member groundwater has no defined dominance, but has notably high magnesium values (~10 mEq/L), as defined from the piper plots. Waters sampled from this member generally have high cation and anion values, with slightly decreasing values to the east of the mining area. The Mt. McRaes Shale groundwater has general sulfate dominance (~10 mEq/L), due to the high sulphur content of the shale unit. Stiff plots show spatial consistency throughout the mine area, except for PZ05SSIX2B, but this could be from contamination from the underlying Mt Sylvia Formation. The Wittenoom Formation has generally low cation/anion values, with a chloride dominance at the PZ17 locality, south of Section Six. The Mt. Sylvia Formation also has low cation/anion values with no particular dominance, displaying a similar stiff plot signature to that of the Mt. McRae Shale groundwater. The Marra Mamba Iron Formation displays anomalously high cation/anion values (20-80 mEq/L in Mg + SO4). As vast amounts of mineralised marthite-geothite ore are present close to the sampling, high values should be expected due to increased permeability and reactions with the regional groundwater system. 92 Chapter 5: Hydrochemistry Mt. Tom Price Stiff Plots Screened Geological Units: Dales Gorge Member Wittenoom Formation Mt. McRae Shale Mt. Sylvia Formation Marra Mamba Iron Formation Na + K Cl Na + K Cl Na + K Cl Ca Ca HCO Ca HCO 3 HCO 3 3 Mg Mg Mg SO 4 SO4 SO4 Fe CO 3 Fe CO3 Fe CO3 mEq /l mEq/l mEq/l mEq/l mEq/l mEq/l 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 Depth Depth NDDW02 ? Depth SEP (DB3) 610.47 mRL WB05SSIX0001 585.26 mRL MM01 622.64 mRL PZ06WEP01 PZ05SSIX2A 577.00 mRL ? Section 6 Pond PZ05SSIX2B 563.51 mRL PZ16 602.79 mRL (S/Pipe) 645 mRL (Pit Lake) WB06NTD1 WEP03 PZ17 (2006) 559.90 mRL 604.90 mRL* 670.36 mRL WEP04 WB05NTD1 698.78 mRL 612.25 mRL WB03SEP 0001 555.30 mRL* WEP05 mEq/l mEq/l WB05SEP02 80 40 0 40 80 570.72 mRL 708.01 mRL MM02 673.53 mRL (S/Pipe) 20 10 0 10 20 20 10 0 10 20 Figure 5.7: Mt. Tom Price stiff plots (February-August 2007 EDMS data). 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 93 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 20 10 0 10 20 Chapter 5: Hydrochemistry Figure 5.8: Mt. Tom Price stiff plot locations (January – June 2007 EDMS data). data). EDMS 2007 June – (January locations plot stiff Price Tom Mt. 5.8: Figure 94 Chapter 5: Hydrochemistry 5.4.3 Discussion Results suggest that groundwater quality is highly dependent upon geological hosts. Clear patterns are defined in the stiff plots, with anomalously high values seen in the Marra Mamba Iron Formation. Therefore we can expect higher initial levels of saline, magnesium and sulphur rich waters in the Marra Mamba West as compared to other pit lakes in the mine area. There appears to be no obvious spatial variation apparent in the results available. Chemical signatures seem to be characteristically similar throughout, displaying variations associated with geological variances and possible disturbance through mining activities. Using the results presented, estimations can be used to give an indication on long term final pit lake quality. As the dominant water quality driver for is evaporative concentration, current water quality values can be used as a reference in the estimation of long term, final pit lake conditions. 5.5 FUTURE PREDICTIONS Using these results collected from recent studies, as well as considering previous estimates of final pit water quality conditions, a revised final pit water quality prediction can be made. As final pit water quality is highly influenced by final water levels, previous predicted levels from the EWL Sciences report should be compared to final predicted water levels generated from the MTPPGM. Estimations on 100 year pit lake salinity have been calculated using the constants derived from the Aquaterra 2002 pit lake quality report (Hall, 2002). As the pit walls comprise of tight BIF and shale, throughflow is minimal and therefore the dominant water quality control is evaporative concentration (Hall, 2002). Marra Mamba Iron Formation groundwater tends to have high salinity values (Figure 5.7), so was incorporated into the model accordingly (Table 5.4). The Mt. McRae Shale contains large amounts of acid forming materials and is likely to generate alarmingly large high levels of acidity in the pit lake due to oxidation reactions with the exposed shale above the pit lake. Of the pit voids studied in this thesis, the South East Prongs pit and Section Six pits are the only areas where large amounts of acid forming shale 95 Chapter 5: Hydrochemistry are likely to contribute to fit pit water quality. The area of shale above the SEP and SSIX pit lake was calculated from aerial photography images. Acidity values were calculated according to the Southern Ridge EWL Sciences groundwater quality report (Jones, 2002), using similar constants to produce comparable results (Table 5.5). SEP values suggest lower long term acidity conditions (Acid/Alkali ratio of 20) compared to the Southern Ridge results (Table 5.2). This is due to the dilution of H2SO4 in the large, deep SEP pit lake. Section 6 however has a much lower volume, with the potential for low pH values in this final pit lake (Acid/Alkali ratio of 500). Table 5.4: Final pit lake dimensions and conditions Predicted Area of Mt. Predicted Pit Time Taken For Pit Lake Salinity After McRae Shale Void Water Full Recovery Volume (ML) 100 Years Above Pit Lake Level (mRL) (Years) 2 (mg/L) (m ) 710 120 19,690 4000 0 North Deposit 685 115 17,030 5000 140,000 South East Prongs 635 96 1,180 10000 0 Marra Mamba West 640 93 100 4000 14,000 Section Six 700 98 NA NA 0 West Pits Table 5.5: Final pit lake conditions according to pit lake levels where Mt. McRae Shale is present H2SO4 equiv Area of MCS Water Vol Total CaCO3 100y Volume H2SO4 Pits RL (m) CaCO3 above pit lake acid/alkali (m3) (tonnes) of MCS (m3) (tonnes) (tonnes) (m2) SEP 685 17,030,000 2,200 2,000 140,000 222,000 40,000 20 SSIX 640 100 0.02 0.02 14,000 20,000 10 500 Results indicate that final pit lake waters will be classified as Brine waters (>100,000 mg/L), as predicted from similar results obtained from the EWL Sciences report (Jones, 2002). This is caused by high evaporation rates and low throughflow in the pit voids. As the pit fills with water, it can be assumed that low throughflow rates will decrease, creating mixing buckets for which water quality is dominantly influenced by evaporation concentration. The South East Prongs and Section Six final pit lakes are likely to have high levels of acidity due to the large area of acid forming Mt. McRae Shale directly above the pit lake. These reactions are likely to slow down over time as the water table recovers due to low dissolved oxygen values in a ‘more stagnant’ pit lake. As oxidation reactions decrease, Aluminium 96 Chapter 5: Hydrochemistry values are expected to rise from an initial high value of 360 mg/L (Table 5.3) (pers. comm., T. Horton, 2008). 5.6 SYNTHESIS Mt. Tom Price hydrochemistry data is sampled regularly by onsite hydrogeologists and environmental teams. Due to the sulphur rich, acid forming Mt. McRae Shale, regular monitoring of pit and groundwater is essential. Previous work has outlined the acid forming nature of the Mt. McRae Shale and has predicted likely groundwater quality of final pit levels for several pits in the mine area, approximated from previous groundwater model outputs. Results from this study indicate minor spatial variance in groundwater conditions over the mine site, with only small fluctuations over the period of analysis (1999-2007). Groundwater sampled from pumped bores is highly influenced by geological hosts, with clearly defined hydrochemical signatures approximated for each screened geological unit. Final pit lake water quality was estimated using final pit levels and recovery rates approximated from the MTPPGM, along with historical data and previous groundwater quality reports. Pit lake water quality is dominantly driven by evaporation concentration, caused by high evaporation rates and low throughflow. Pit waters are expected to be brine waters, with high levels of acidity values occurring in the South East Prongs and Section Six pits. 97 REFERENCES Anderson, M.P., and Woessner, W.W., 1992, Applied Groundwater Modelling - Simulation of Flow and Advective Transport, Academic Press, 381 p Blake, T.S., and Barley, M.E., 1992, Tectonic evolution of the Late Archaean to Early Proterozoic Mount Bruce Megasequence Set, Western Australia, Tectonics vol. 11, 1415- 1425 p Chiang, W., 2005, 3D-Groundwater Modeling with PMWIN, Second Edition - A Simulation system for Modeling Groundwater Flow and Transport Processes, Springer, 411 p Doherty, J., 2000, PEST - Model-Independent Parameter Estimation, Fifth Edition, Watermark Numerical Computing, 249 p Drever, J. I., 1997, The Geochemistry of Natural Waters: Surface and Groundwaters, 3rd Edition, Prentice-Hall inc, New Jersey, 436 p Druzynski, A., and Hall, J., 2004, SSIX Groundwater Modelling - Simulation of Discharge of Excess Flood Waters (R075b.pdf), Hamersley Pty Ltd/Aquaterra, 18 p Etheridge, M., Henley, J., Williams, P., 1996, Controls on the Localisation of Primary Hypogene Iron Enrichment (PIE) in the Hamersley Basin, Circulation Hamersley Iron Pty. Ltd. EHW Pty. Ltd. Canberra, l-64 p Fetter, C. W., 2000, Applied Hydrogeology, Fourth Edition, Prentice Hall, 598 p Gilhome, W.R., 1975, Mount Tom Price Iron Orebody, Hamersley Iron Province, Economic Geology of Australia and Papua New Guinea 1, Metals. Monograph No. 5. Ed, C.L. Knight, 892-898 p Hall, J., 2002, Hamersley Pit Water Quality (Water Quality StrnRG-SEP.doc), ELW/Aquaterra, 10 p Hall, J., and Rozlapa, K., 2002, Impact of Final Voids on Groundwater Flows and Quality (Impact of final voids SEP-NTD-STRG.doc), Hamersley Iron Pty Ltd/Aquaterra, 21 p Hamersley Exploration Ltd, 1972, Tom Price Geological Survey (GDSR1025.pdf), Hamersley Exploration Ltd, 91 p Hamersley Iron Pty Ltd, 2000, The Geology and Mineralogy of the Hamersley Provence Iron Ores – The Green Book, Hamersley Iron Pty Ltd, 99 p 104 Harmsworth, R.A., Kneeshaw, M,, Morris, R.C., Robinson, C.J and Shrivastava, PK., 1990, BIF-Derived Iron Ores of the Hamersley Province. In: Geology of the Mineral Deposits of Australia and Papua New Guinea, Monograph No.14. ed. F.E.Hughes, 617-642 p Hill, M.C. and Tiedeman, C.R., 2007, Effective groundwater model calibration: with analysis of data, sensitivities, predictions, and uncertainty, Hoboken, N.J. : Wiley-Interscience, 455 p Jones, D, 2002, Memorandum - Future Water Quality of Hamersley Iron Pits, EWL Sciences, 10 p Krapez, B., 1997, Sequence-stratigraphic concepts applied to the identification of depositional basins and global tectonic cycles, Australian Journal of Earth Sciences, vol. 44, l-36 p Kresic, N., 1997, Quantitave Solutions in Hydrogeology and Groundwater Modeling, Lewis Publishers, 461 p McDonald, M.G., and Harbaugh, A.W., 1988, A modular three-dimensional finite-difference ground-water flow model: U.S, Geological Survey Techniques of Water-Resources Investigations, 586 p Preston, M., 1994, Tom Price Hydrogeological Studies (RTIO-HSE-0016705.pdf), Ultramafics Pty Ltd, 76 p Rathbone, S., 2005, A groundwater model of the 4WEST and 4EAST pits, incorporating the Seven Mile, Pirraburdoo Creeks and Ratty Springs. Paraburdoo - Western Australia, MSc Thesis, UWA/Rio Tinto Iron Ore, 77 p Rozlapa, K., 2004, South East Prongs Dewatering (059b.pdf), Aquaterra/Hamersley Iron Pty Ltd, 9 p Rozlapa, K., and Hall, J., 1999, North Deposit Recalibration (096a.pdf), Aquaterra, 10 p Trendall, A.F. and Blockely. J.G., 1970, The Iron Formations of the Precambrian Hamersley Group Western Australia - with special reference to the associated Crocidolite, Geological Survey of Western Australia, 1-366 p Trendall, A.F., Nelson. D.R., De Laeter, J.R., and Hassler, S.W., 1998. Precise zircon U-Pb ages from the Marra Mamba Iron Formation and Wittenoom Formation, Hamersley Group, Western Australia. Aust. Jour. Earth. Sci. vol. 45, 137-142 p Tyler, I.M., and Thorne, A.M., 1990, The northern margin of the Capricorn Orogen, Western Australia – an example of an Early Proterozoic collision zone, Journal of Structural Geology, volume 138, 108 p http://www.riotintoironore.com, 2008, Rio Tinto Iron Ore. 105 Appendix 1: 2007 Water Balance Appendix 2: Geological Cross-sections le). ca ater model (50 m vertical s ng to each layer in the groundw is always 520 mRL). Formation - Lower Wittenoom Marra MambaFormation Jeerinah Formation Fault Zone Mineralised BIF Dolerite Dike Fault in Figure 3.2 (looking west). ents interpreted geology accordi (i.e. base of layer 6) Detritus Joffre Member Whaleback Shale Formation Dales Gorge Member Mt McRae Shale Mt Sylvia Formation Formation - Upper Wittenoom Cross Section localities found Mt Tom Price Geological Cross-Sections Mt Tom Price Geological Key: Each rectangular block repres NB: Base of rectangular blocks Cross-section 25 (8990 E) E) (8990 25 Cross-section Appendix 2: Geological Cross-sections Cross-section 2 (9900 E) E) (10320 11 Cross-section Cross-section 1 (10550 E) E) (10940 12 Cross-section Appendix 2: Geological Cross-sections Cross-section 14 (11300 E) E) (11300 14 Cross-section E) (11650 15 Cross-section Cross-section 3 (12330 E) Cross-section 9 (12700 E) Appendix 2: Geological Cross-sections 14: Cross-section 1 Cross-section 10 (13500 E) E) (13500 10 Cross-section E) (13970 16 Cross-section E) (14500 17 Cross-section E) (15020 18 Cross-section Appendix 2: Geological Cross-sections Cross-section 19 (15720 E) E) (15720 19 Cross-section E) (16230 20 Cross-section E) (16700 21 Cross-section Cross-section 8 (17260 E) Cross-section 7 (17630 E) Appendix 2: Geological Cross-sections Cross-section 6 (18030 E) Cross-section 5 (18600 E) Cross-section 4 (19080 E) Appendix 3: Previous Hydrogeological Parameters Material / Formation kH Sy S Alluvium (undefined) 0.1 - 10 0.01 – 0.2 TD3 – Oakover Fm TD2 – CID (Robe Pisolite) 2-155 0.025-0.05 1.0x10-4 TD2 – Calcrete 5 - 50 0.02 – 0.05 1.0x10-4 TD1 – clay / pisoliths Fresh Boolgeeda Fm / Wongarra Volcs 0.0008-0.1 5x10-5 1x10-5 Fractured Boolgeeda Fm / Wongarra Volcs 5-155 0.025-0.05 0.03 – 5x10-4 Weeli Wolli Fm 0.1 0.001 1.0x10-5 Brockman Iron Fm (unmineralised BIF) 1x10-4 - 0.01 0.001 1.0x10-5 Mineralised Joffre (Mt Whaleback) 1 – 5 0.04 – 0.06 2.0x10-4 Whaleback Shale (Mt Whaleback) 0.01 - 0.02 0.01 2.0x10-4 Mineralised Dales Gorge (Mt Whaleback) 1 – 5 0.04 – 0.06 2.0x10-4 Mt McRae Shale / Mt Sylvia Shale 0.01 2.0x10-4 Bruno’s Band (at Mt whaleback) 17 0.01 2.0x10-4 Mt Sylvia Fm / Bee Gorge Member 0.1 0.001 1.8x10-5 Paraburdoo Dolomite 5 - 8 0.001 - 0.005 3.1x10-4 West Angelas Shale 0.01 – 0.5 0.001 1.0x10-5 - 2.0x10-4 Mineralised West Angelas Shale (WA) 0.4 0.01 – 0.03 Marra Mamba Fm (unmineralised BIF) 1x10-4 0.001 1.0x10-5 Marra Mamba Ore (MAC, HD, WA) 2.8 - 8 0.01 - 0.1 9.0x10-4 - 2.7x10-3 Marra Mamba Sub-grade Ore (MAC, HD, WA) 0.001 – 0.004 0.001 – 0.03 1.0x10-5 - 2.0x10-4 Jeerinah Lineaments 0.3 0.0005 Jeerinah Fm 0.03 0.03 2.0x10-4 Source: Rio Tinto Hydrogeological Database Appendix 4: Geological Templates Geological Model Templates for Groundwater Model Input Key: Jeerinah Formation Marra Mamba Formation Wittenoom Formation Mt Slyvia Formation Mt McRaes Shale Dales Gorge Member Whaleback Shale Joffre Formation Pit Void NB: variance in patterns due hatching difficulties in AUTOCAD. Irregular white patterns in Layers 1-4 represent a topographical surface below the layer mRL. Layer 1-2 geological template (820 - 770 mRL) Appendix 4: Geological Templates Jeerinah Layer 3-4 geological template (770 - 720 mRL) Jeerinah Layer 5-6 geological template (720 - 670 mRL) Appendix 4: Geological Templates Jeerinah Layer 7-8 geological template (670 - 620 mRL) Jeerinah Layer 9-10 geological template (620 - 570 mRL) Appendix 4: Geological Templates Jeerinah Layer 11-12 geological template (570 - 520 mRL) Appendix 5: Initial Hydraulic Heads Initial WL Depth Name Easting Northing mRL Pit Area Top mRL Depth m mRL Geology Type Installed Status DB1 16190.7 10652.7 654.8 SEP 735.20 120 615.20 ? Bore Piezo ? Historical DB2 15553 10471.8 660.21 SEP 676.08 ? ? ? Bore Piezo ? Historical DB3 15552.56 10471.07 659.27 SEP 661.22 80 581.22 ? Bore Piezo 1-Feb-04 Historical DB4 10464 11923.1 670.63 NTD 690.28 96 594.28 ? Bore Piezo ? Historical ND143(PZ33) 10674 11934.7 695.87 NTD 709.57 ? ? ? S/Pipe ? Historical ND147(PZ34) 10613.9 11830.9 676.52 NTD 699.72 ? ? ? S/Pipe ? Historical ND183(PZ36) 10573.9 11992.8 678 NTD 710.53 ? ? ? S/Pipe ? Historical ND189(PZ37) 10272.5 11825.2 674.97 NTD 699.36 ? ? ? S/Pipe ? Historical ND196(PZ38) 10657.2 12013.9 685.81 NTD 715.11 ? ? ? S/Pipe ? Historical ND69(PZ32) 10474.2 11707.8 674.35 NTD 697.15 ? ? ? S/Pipe ? Historical NDP1(PZ39) 10464.1 11923 677.37 NTD 699.30 ? ? ? S/Pipe ? Historical OBW 15573.4 10485.9 656.5 SEP 679.60 ? ? ? S/Pipe ? Historical OBW 15593.6 10500.4 659 SEP 679.90 ? ? ? S/Pipe ? Historical OBW1 12929.1 10760.4 780.47 CTR 833.17 120 713.17 MCS S/Pipe 28-Aug-94 Historical OBW2(WPIT-N) 12477.8 11098.8 742 WPIT 780.23 150 630.23 ? S/Pipe 29-Aug-94 Historical OBW4 15609.8 10516.8 659.88 SEP 679.90 ? ? DG3 S/Pipe ? Historical PN690/3 15555.1 10374.2 676.35 SYN 694.70 ? ? ? P'matic ? Historical PN690/4 15555.1 10423.8 676.4 SYN 694.70 ? ? ? P'matic ? Historical PN690/5 15529 10327.5 676.39 SEP 691.20 ? ? ? P'matic ? Historical PN690/6 15529 10328.2 676.44 SEP 691.30 ? ? ? P'matic ? Historical PN720/2 16155.7 10552 676.34 SEP 721.20 ? ? ? P'matic ? Historical PX26 13929 9250 632.14 SSIX 720.95 120 600.95 DG1 + FWZ S/Pipe ? Historical PZ10 15158.7 10613 674 SYN 765.40 120 645.40 MCS S/Pipe 6-Sep-94 Historical PZ11 15299.8 10179.8 685 SYN 765.00 140 625.00 MTS S/Pipe 5-Sep-94 Historical PZ12 14849.6 10559.9 727.46 SYN 795.96 120 675.96 MCS S/Pipe 24-Aug-94 Historical PZ16 (XROADS) 15921.8 9682.4 669 SSIX 767.79 165.00 602.79 ? S/Pipe ? Historical PZ17 12995.1 8550.9 631.5 SSIX 709.90 150 559.90 ? Piezo ? Current PZ18 14350.9 8556.7 631.5 SSIX 721.99 175 546.99 ? S/Pipe ? Historical PZ19 13850 9030 633.36 SSIX 711.35 100 611.35 MCS S/Pipe ?/9/94 Historical PZ20 14327 9300 646.5 SSIX 737.35 160 577.35 MCS S/Pipe ? Historical PZ21 13902.9 10639.7 742.39 CTR 834.89 150 684.89 MCS S/Pipe ? Historical PZ22(WPIT) 12498.8 11001.1 773.45 WPIT 779.81 100 679.81 ? S/Pipe ? Historical PZ23(WPIT-E) 12961.5 10976 760.68 WPIT 795.55 140 655.55 ? S/Pipe ? Historical PZ24 13114.2 10683.2 809.38 CTR 815.75 45 770.75 FWZ + MCS S/Pipe ? Historical PZ25 13087.6 10683.4 808.81 CTR 818.08 ? ? FWZ + MCS S/Pipe ? Historical PZ27 14001.5 8950.8 629.5 SSIX 720.50 120 600.50 MCS S/Pipe 9-Sep-94 Historical PZ35 10232.86 11883.3 675.96 NTD 701.63 71.9 629.73 ? Piezo ? Current PZ5 16705 10685 701.87 SEP 750.00 ? ? ? S/Pipe ? Historical PZ5a 16189.9 10472.3 656.84 SEP 705.40 74 631.40 DG S/Pipe 2-Sep-94 Historical PZ6 16710 10685 679 SEP 751.40 160 591.40 ? S/Pipe ? Historical PZ61 16100.3 10641.3 672.76 SEP 733.90 ? ? ? S/Pipe ? Historical PZ62 16099.9 10569.9 672.37 SEP 720.40 ? ? ? S/Pipe ? Historical PZ63 16219.2 10657.5 676.72 SEP 735.00 ? ? ? S/Pipe ? Historical PZ64 16155.6 10659.1 674.07 SEP 734.40 ? ? ? S/Pipe ? Historical PZ69 15783.6 10503.8 659.5 SEP 690.90 ? ? ? S/Pipe 23-Sep-94 Historical PZ6a 15702.8 10555.6 669 SEP 691.32 100 591.32 MCS + MTS S/Pipe 6-Sep-94 Destroyed PZ70 15841.994 10517.08 657.1 SEP 689.97 ? ? ? S/Pipe ? Historical PZ7a 15783.6 10503.7 659.7 SEP 690.80 64 626.80 DG S/Pipe ? Historical PZ8 16120.2 10201.3 671.5 SEP 764.90 170 594.90 MCS dry S/Pipe ? Historical S90/10 15714.6 10630.6 663.8 SEP 736.76 ? ? ? P'matic ? Historical S90/13 16081.9 10665.3 672.79 SEP 750.70 ? ? ? P'matic ? Historical S90/8 16091 10671 673.23 SEP 749.90 ? ? ? S/Pipe ? Historical SP690/1 15555 10448.5 676.38 SEP 690.30 ? ? ? S/Pipe ? Historical SP690/2 15555.1 10374.2 676.25 SEP 690.20 ? ? ? S/Pipe ? Historical SP720/1 16155.3 10601.2 676.47 SEP 720.80 ? ? ? S/Pipe ? Historical SP720/3 16138.5 10439.2 676.48 SEP 722.20 ? ? ? S/Pipe ? Historical P41* 10073.63* 7649.1* 631.46* SSEV ? ? ? ? ? ? ? P64* 12997.97* 15958* 650* NA NA NA NA NA NA NA NA P65* 17812.96* 13462* 720* NA NA NA NA NA NA NA NA Appendix 5: Initial Hydraulic Heads Initial hydraulic heads: Surfer generated (contours in mRL) Initial hydraulic heads: Steady state model output (contours in mRL) Appendix 6: Groundwater Model Layers – Hydraulic Conductivity PMWIN Pro Mt Tom Price Groundwater Model Layers – Horizontal (KH) and Vertical Conductivity Parameters (KV) Key: KH (m/day) KV (m/day) Jeerinah Formation 0.03 0.003 Marra Mamba Formation 0.0001 0.0001 Wittenoom Formation 0.1 0.01 Mt Slyvia Formation 2 1.7 Mt McRaes Shale 0.01 0.001 Dales Gorge Member 0.005 0.0005 Whaleback Shale 0.008 0.0008 Joffre Formation 0.001 0.0001 Pit Void 100 100 Fault Zone 1 1 Inactive Cells NB: Other zones visible represent areas that were added to lower conductivity contrasts. Layer 1 Appendix 6: Groundwater Model Layers – Hydraulic Conductivity Layer 2 Layer 3 Appendix 6: Groundwater Model Layers – Hydraulic Conductivity Layer 4 Layer 6 Appendix 6: Groundwater Model Layers – Hydraulic Conductivity Layer 7 Layer 8 Appendix 6: Groundwater Model Layers – Hydraulic Conductivity Layer 9 Layer 10 Appendix 6: Groundwater Model Layers – Hydraulic Conductivity Layer 10 Layer 12 Appendix 7: Model Layers – Specific Yield PMWIN Pro Mt Tom Price Model Layers – Specific Yield (SY) Key: Jeerinah Formation Whaleback Shale Pit Void Other Inactive Cells Layer 2 Appendix 7: Model Layers – Specific Yield Layer 3 Layer 4 Appendix 7: Model Layers – Specific Yield Layer 5 Layer 6 Appendix 7: Model Layers – Specific Yield Layer 7 Layer 8 Appendix 7: Model Layers – Specific Yield Layer 9 & 10 Layer 11 & 12 Appendix 8: Model Layers - Storage PMWIN Pro Model Layers – Storage Coefficient (SC) Key: Jeerinah, Wittenoom & Mt McRaes Shale Mt Slyvia Formation Marra Mamba Formation, Dales Gorge Member Pit Void Other Inactive Cells Layer 2 Appendix 8: Model Layers - Storage Layer 3 Layer 4 Appendix 8: Model Layers - Storage Layer 5 Layer 6 Appendix 8: Model Layers - Storage Layer 7 Layer 8 Appendix 8: Model Layers - Storage Layer 9 & 10 Layer 11 & 12 Appendix 9: Water Pump List Used In MTPGM Name WB05NTD1 WB03NTD1 WB06NTD1 WB06NTD2 WB05SEP1 WB05SEP2 WB06SEP1 WB06SEP2 WB05SSEV01 WB04STR2 WB03SEP1 WB05SSIX1 WB03NTD‐1 DB1 DB2 DB3 SSIX Discharge Easting 10393.512 10261.753 10196.79 10261.753 15974.022 15904.73 16100.289 15555.604 12652.157 13739.798 16142.399 13926.781 10261.753 16191 15553 15847 14145 Northing 12260.524 11927.655 12366.9 11927.655 10546.91 10446.25 10511.814 10457.952 7919.5 10424.614 10539.485 9196.36 11927.66 10653 10472 10514 9245 Top mRL 685.25 688.39 674.9 684.34 631.34 615 614.84 615.33 698.09 795 615.30 660.06 688.39 735.2 680.5 690.47 630 Depth m 73 76 ? 142.2 74.5 60 ? 78.2 137 78 ? 74.8 106 120 ? 80 NA Depth mRL 612.25 612.39 ? 542.14 556.84 555 ? 537.13 561.09 717 ? 585.26 582.39 615.20 ? 610.47 NA Layer 8 8 9 11 11 11 11 11 10 5 11 10 10 8 11 8 8 Row 102 115 97 115 161 165 162 164 238 165 161 204 115 155 163 162 202 Column 88 83 80 83 295 292 300 278 162 205 302 213 83 304 278 290 222 Bore Number 1 2 3 4 5 6 7 8 9 10 13 14 15 25 26 27 28 Date Stress Period Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day Avg kL/day 1994 1 100 15 50 15 10 1995 2 305 30 100 30 25 1996 3 305 30 100 30 25 1997 4 305 30 100 30 25 1998 5 305 30 100 30 25 1999 6 305 30 100 30 25 2000 7 305 30 100 30 25 2001 8 305 30 100 30 25 2002 9 305 30 100 30 25 2003 10 305 30 100 30 25 Jan‐04 11 150 11 1000 Jun‐04 12 32 1166 2900 Jul‐04 13 32 666 1900 Aug‐04 14 32 333 1170 Sep‐04 15 32 805 Oct‐04 16 50 890 Nov‐04 17 0 860 Dec‐04 18 1666 500 1330 Jan‐05 19 2333 200 2685 55* Feb‐05 20 2266 500 2300 339* Mar‐05 21 2333 116 980 2230 227* Apr‐05 22 1766 153 620 1800 700* May‐05 23 333 1666 166 690 1640 500 918* Jun‐05 24 333 166 66 0 175 700 796* Jul‐05 25 666 1166 0 1170 280 401* Aug‐05 26 666 1400 650 1550 750 732* Sep‐05 27 666 1333 258 133 390 300 470 604* Oct‐05 28 1333 1100 276 0 66 0 1360 0 102* Nov‐05 29 1433 666 423 266 33 0 550 0 458* Dec‐05 30 1833 1833 2666 366 433 330 200 33 550 1830 150 1713* Jan‐06 31 1000 1266 2000 333 766 600 733 166 650 1400 780 2258* Feb‐06 32 1666 1500 1600 366 666 600 666 233 360 740 450 2596* Mar‐06 33 333 1066 1000 400 500 600 666 466 1610* Apr‐06 34 333 700 660 333 866 600 666 100 1313* May‐06 35 1833 666 660 400 866 800 666 1749* Jun‐06 36 2333 1333 333 400 766 500 666 2357* Jul‐06 37 2066 1466 1333 366 766 200 533 979* Aug‐06 38 2166 1366 1333 333 633 100 500 1341* Sep‐06 39 2333 1033 1166 200 466 50 266 729* Oct‐06 40 2333 400 1166 333 100 500 266 1053* Nov‐06 41 2333 2666 1166 166 600 433 433 1234* Dec‐06 42 2166 2666 666 333 333 466 633 300 1673* Jan‐07 43 2166 3600 433 166 0 433 500 543* Feb‐07 44 833 2833 366 100 333 466 400 1067* Mar‐07 45 833 2666 333 166 566 133 400 865* Apr‐07 46 2666 433 333 466 200 433 978* May‐07 47 2633 166 200 366 266 1176* Jun‐07 48 2633 166 200 366 266 983* * Represents positive pumping wells (i.e. discharge) Appendix 10: Observation Bore List - MTPGM Top Name Easting Northing (mRL) Depth (m) Depth (mRL) Layer Type Geology DB2 15553 10472 680.5 70 610.50 10 Bore Piezo DG DB3 15551.83 10470.21 690.47 80 610.47 10 Bore Piezo DG MM1 13672.25 8186.54 712.64 90.00 622.64 NA (8) S/Pipe MM NW4 16200.72 10819.05 751.09 142.85 608.24 8 Piezo WD NW5 15606.53 10810.59 751.69 129.80 621.89 8 Piezo WD P00NW001 10698.14 11656.67 718.38 102 616.38 10 Piezo MCS PZ03SEP3 16376.1 10378.57 734.29 174 560.29 11 Piezo MTS PZ03SEP4 16670.32 10479.96 768.95 190 578.95 10 Piezo WD/MTS PZ04SEP1 16020.58 10477.21 646.40 54 592.40 10 Piezo DG/FWZ PZ05MM01 11962.67 7625.44 697.07 78.80 618.27 10 Piezo MM PZ05SEP7 15431.93 10312.48 690.05 109.9 580.15 10 Piezo WD PZ05SIX1 13866.06 9179.6 661.60 78.15 583.45 10 Piezo MCS PZ17 12995.1 8550.9 709.90 150 559.90 10 Piezo MCS PZ18 14350.9 8556.7 721.99 175 546.99 NA (11) S/Pipe MM PZ35 10232.86 11883.3 701.63 71.9 629.73 8 Piezo MCS PZ3NTD13 10654.44 11738.97 711.96 68 643.96 8 Piezo DG PZ3SEP1 16261.79 10907.22 794.67 130 664.67 7 Piezo MTS PZ5MM2 17642.4 10615.63 696.08 90.05 606.03 9 Piezo MM PZ5MM3 11962.67 7625.44 720.16 94.00 626.16 8 Piezo MM PZ5SIX3A 14093.31 9038.65 677.959 107.1 570.86 10 Piezo MTS PZ6NTD2B 10202.57 12352.75 675.38 150 525.38 12 Piezo MCS PZ6NTD7 10342.65 11768.24 690.341 100 590.34 10 Piezo DG PZ6SEP15 15904.67 10450.6 616.15 80 536.15 11 Piezo DG PZ8 16120.2 10201.3 764.95 170 594.95 9 S/Pipe MTS/WD SSEV1 9998.8 7560.47 699.89 81 618.89 8 Piezo DG WB03NTD1 10261.75 11927.66 688.39 106 582.39 9 Bore Piezo DG WEP3 11826.43 10892.23 750.36 80 670.36 6 Piezo MCS Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 11: Initial Model Hydrographs Appendix 12: Initial PEST Optimisation Results OPTIMISATION ITERATION NO. : 1 Model calls so far : 1 Starting phi for this iteration: 3.13517E+05 Parameter "hk13" has no effect on observations. Parameter "vk9" has no effect on observations. Parameter "vk18" has no effect on observations. Parameter "sc19" has no effect on observations. Lambda = 10.000 -----> Phi = 2.64895E+05 ( 0.845 of starting phi) Lambda = 5.0000 -----> Phi = 2.76495E+05 ( 0.882 of starting phi) Lambda = 20.000 -----> Phi = 2.70905E+05 ( 0.864 of starting phi) No more lambdas: phi rising Lowest phi this iteration: 2.64895E+05 Current parameter values Previous parameter values hk1 0.734974 hk1 1.00000 hk2 0.752222 hk2 2.00000 hk3 68.2525 hk3 100.000 hk4 2.063699E-03 hk4 1.000000E-02 hk5 1.248760E-02 hk5 5.000000E-03 hk6 2.123658E-02 hk6 3.000000E-02 hk7 5.411571E-03 hk7 8.000000E-03 hk8 1.020671E-03 hk8 1.000000E-03 hk9 6.147161E-02 hk9 8.000000E-02 hk11 7.894003E-02 hk11 5.000000E-02 hk12 4.163485E-04 hk12 5.000000E-04 hk13 1.000000E-05 hk13 1.000000E-05 hk15 6.225402E-05 hk15 5.000000E-05 hk19 0.115790 hk19 0.100000 hk20 1.000000E-03 hk20 1.000000E-04 hk21 1.037056E-04 hk21 1.000000E-04 vk1 0.738714 vk1 1.00000 vk2 1.66428 vk2 2.00000 vk3 110.984 vk3 100.000 vk4 1.201554E-02 vk4 1.000000E-02 vk5 4.220098E-03 vk5 5.000000E-03 vk6 3.461830E-02 vk6 3.000000E-02 vk7 4.669191E-03 vk7 8.000000E-03 vk8 2.892732E-03 vk8 1.000000E-03 vk9 8.000000E-02 vk9 8.000000E-02 vk11 5.197300E-02 vk11 5.000000E-02 vk12 5.180785E-04 vk12 5.000000E-04 vk13 1.113057E-05 vk13 1.000000E-05 vk14 2.960063E-03 vk14 2.000000E-03 vk15 4.475450E-05 vk15 5.000000E-05 vk16 0.523270 vk16 0.500000 vk17 1.13625 vk17 1.70000 vk18 2.000000E-04 vk18 2.000000E-04 vk19 6.844843E-02 vk19 0.100000 vk20 2.408287E-04 vk20 1.000000E-04 sy4 9.711583E-02 sy4 1.000000E-02 sy6 1.944440E-02 sy6 3.000000E-02 sy8 8.658120E-04 sy8 1.000000E-03 sy22 4.07586 sy22 0.990000 sc8 1.132843E-03 sc8 1.000000E-03 sc13 2.376150E-05 sc13 1.000000E-05 sc18 3.468632E-04 sc18 2.000000E-04 sc19 2.000000E-04 sc19 2.000000E-04 sc22 0.692880 sc22 0.990000 Appendix 12: Initial PEST Optimisation Results sc23 1.253764E-05 sc23 1.800000E-05 Maximum factor change: 10.00 ["hk20"] Maximum relative change: 9.000 ["hk20"] Optimisation complete: optimisation iteration limit of 1 realized. Total model calls: 49 The model has been run one final time using best parameters. Thus all model input files contain best parameter values, and model output files contain model results based on these parameters. OPTIMISATION RESULTS Covariance matrix and parameter confidence intervals cannot be determined:- Normal matrix nearly singular; cannot be inverted. Parameters -----> Parameter Estimated value hk1 0.734974 hk2 0.752222 hk3 68.2525 hk4 2.063699E-03 hk5 1.248760E-02 hk6 2.123658E-02 hk7 5.411571E-03 hk8 1.020671E-03 hk9 6.147161E-02 hk11 7.894003E-02 hk12 4.163485E-04 hk13 1.000000E-05 hk15 6.225402E-05 hk19 0.115790 hk20 1.000000E-03 hk21 1.037056E-04 vk1 0.738714 vk2 1.66428 vk3 110.984 vk4 1.201554E-02 vk5 4.220098E-03 vk6 3.461830E-02 vk7 4.669191E-03 vk8 2.892732E-03 vk9 8.000000E-02 vk11 5.197300E-02 vk12 5.180785E-04 vk13 1.113057E-05 vk14 2.960063E-03 vk15 4.475450E-05 vk16 0.523270 vk17 1.13625 vk18 2.000000E-04 vk19 6.844843E-02 vk20 2.408287E-04 sy4 9.711583E-02 sy6 1.944440E-02 sy8 8.658120E-04 sy22 4.07586 sc8 1.132843E-03 sc13 2.376150E-05 sc18 3.468632E-04 sc19 2.000000E-04 sc22 0.692880 sc23 1.253764E-05 Appendix 13: Initial PEST Super Parameter Results OPTIMISATION RECORD INITIAL CONDITIONS: Sum of squared weighted residuals (ie phi) = 3.13517E+05 Current parameter values par1 10.0000 par2 10.0000 par3 10.0000 par4 10.0000 par5 10.0000 par6 10.0000 par7 10.0000 par8 10.0000 par9 10.0000 par10 10.0000 OPTIMISATION ITERATION NO. : 1 Model calls so far : 1 Starting phi for this iteration: 3.13517E+05 Lambda = 10.000 -----> Phi = 2.79396E+05 ( 0.891 of starting phi) Lambda = 5.0000 -----> Phi = 2.81166E+05 ( 0.897 of starting phi) Lambda = 20.000 -----> Phi = 2.76705E+05 ( 0.883 of starting phi) No more lambdas: relative phi reduction between lambdas less than 0.0300 Lowest phi this iteration: 2.76705E+05 Current parameter values Previous parameter values par1 10.0460 par1 10.0000 par2 10.1036 par2 10.0000 par3 10.1817 par3 10.0000 par4 10.0392 par4 10.0000 par5 10.2852 par5 10.0000 par6 11.0000 par6 10.0000 par7 10.4378 par7 10.0000 par8 9.16779 par8 10.0000 par9 9.85175 par9 10.0000 par10 9.64801 par10 10.0000 Maximum relative change: 0.1000 ["par6"] OPTIMISATION ITERATION NO. : 2 Model calls so far : 4 Starting phi for this iteration: 2.76705E+05 Lambda = 20.000 -----> Phi = 2.62904E+05 ( 0.950 of starting phi) Lambda = 10.000 -----> Phi = 2.66250E+05 ( 0.962 of starting phi) Lambda = 40.000 -----> Phi = 2.66264E+05 ( 0.962 of starting phi) No more lambdas: phi rising Lowest phi this iteration: 2.62904E+05 Relative phi reduction between optimisation iterations less than 0.1000 Switch to central derivatives calculation Appendix 13: Initial PEST Super Parameter Results Current parameter values Previous parameter values par1 9.56625 par1 10.0460 par2 9.78741 par2 10.1036 par3 10.3428 par3 10.1817 par4 9.73158 par4 10.0392 par5 10.0210 par5 10.2852 par6 10.6510 par6 11.0000 par7 9.81401 par7 10.4378 par8 9.22103 par8 9.16779 par9 10.0075 par9 9.85175 par10 10.5697 par10 9.64801 Maximum relative change: 9.5528E-02 ["par10"] OPTIMISATION ITERATION NO. : 3 Model calls so far : 17 Starting phi for this iteration: 2.62904E+05 Lambda = 20.000 -----> Phi = 2.66401E+05 ( 1.013 times starting phi) Lambda = 10.000 -----> Phi = 2.65797E+05 ( 1.011 times starting phi) No more lambdas: relative phi reduction between lambdas less than 0.0300 Lowest phi this iteration: 2.65797E+05 Current parameter values Previous parameter values par1 9.45903 par1 9.56625 par2 9.66401 par2 9.78741 par3 10.3976 par3 10.3428 par4 9.25548 par4 9.73158 par5 9.91115 par5 10.0210 par6 10.8666 par6 10.6510 par7 9.86041 par7 9.81401 par8 9.16846 par8 9.22103 par9 10.1883 par9 10.0075 par10 10.7095 par10 10.5697 Maximum relative change: 4.8924E-02 ["par4"] OPTIMISATION ITERATION NO. : 4 Model calls so far : 39 Starting phi for this iteration: 2.65797E+05 Lambda = 5.0000 -----> Phi = 2.57518E+05 ( 0.969 of starting phi) Lambda = 2.5000 -----> Phi = 2.57447E+05 ( 0.969 of starting phi) No more lambdas: relative phi reduction between lambdas less than 0.0300 Lowest phi this iteration: 2.57447E+05 Current parameter values Previous parameter values par1 9.15969 par1 9.45903 par2 9.44502 par2 9.66401 par3 10.6366 par3 10.3976 par4 9.88645 par4 9.25548 par5 9.63264 par5 9.91115 par6 10.4432 par6 10.8666 par7 9.29654 par7 9.86041 par8 10.0853 par8 9.16846 par9 9.97427 par9 10.1883 par10 9.74435 par10 10.7095 Appendix 13: Initial PEST Super Parameter Results Maximum relative change: 0.1000 ["par8"] OPTIMISATION ITERATION NO. : 5 Model calls so far : 61 Starting phi for this iteration: 2.57447E+05 Lambda = 1.2500 -----> Phi = 2.53495E+05 ( 0.985 of starting phi) Lambda = 0.62500 -----> Phi = 2.55093E+05 ( 0.991 of starting phi) Lambda = 2.5000 -----> Phi = 2.54433E+05 ( 0.988 of starting phi) No more lambdas: phi rising Lowest phi this iteration: 2.53495E+05 Current parameter values Previous parameter values par1 8.90582 par1 9.15969 par2 8.57828 par2 9.44502 par3 10.2982 par3 10.6366 par4 9.76730 par4 9.88645 par5 9.59438 par5 9.63264 par6 11.0303 par6 10.4432 par7 8.76555 par7 9.29654 par8 9.86738 par8 10.0853 par9 10.1364 par9 9.97427 par10 9.60010 par10 9.74435 Maximum relative change: 9.1767E-02 ["par2"] OPTIMISED PARAMETERS hk1 0.9606978 1.00000 0.00000 hk2 0.1316453 1.00000 0.00000 hk3 111.1274 1.00000 0.00000 hk4 1.0702677E-02 1.00000 0.00000 hk5 8.8240244E-03 1.00000 0.00000 hk6 3.5692512E-02 1.00000 0.00000 hk7 1.4509696E-02 1.00000 0.00000 hk8 1.0435141E-03 1.00000 0.00000 hk9 7.9679320E-02 1.00000 0.00000 hk11 5.3438081E-03 1.00000 0.00000 hk12 9.8863938E-04 1.00000 0.00000 hk13 1.0000000E-05 1.00000 0.00000 hk15 4.9336257E-05 1.00000 0.00000 hk19 1.6579729E-02 1.00000 0.00000 hk20 1.1883760E-04 1.00000 0.00000 hk21 1.0370678E-04 1.00000 0.00000 vk1 0.9874253 1.00000 0.00000 vk2 2.054198 1.00000 0.00000 vk3 100.3448 1.00000 0.00000 vk4 1.2556534E-02 1.00000 0.00000 vk5 2.4560184E-02 1.00000 0.00000 vk6 3.4428104E-02 1.00000 0.00000 vk7 1.3238498E-02 1.00000 0.00000 vk8 1.3013003E-03 1.00000 0.00000 vk9 8.0000000E-02 1.00000 0.00000 vk11 5.2850755E-02 1.00000 0.00000 vk12 1.0129228E-03 1.00000 0.00000 vk13 9.7733854E-06 1.00000 0.00000 vk14 2.3550689E-03 1.00000 0.00000 vk15 4.8515599E-05 1.00000 0.00000 vk16 0.4908742 1.00000 0.00000 Appendix 13: Initial PEST Super Parameter Results vk17 6.989817 1.00000 0.00000 vk18 2.0000000E-04 1.00000 0.00000 vk19 0.1653735 1.00000 0.00000 vk20 1.5571196E-04 1.00000 0.00000 sy4 9.5048308E-03 1.00000 0.00000 sy6 0.1192416 1.00000 0.00000 sy8 9.4149275E-05 1.00000 0.00000 sy22 2.093205 1.00000 0.00000 sc8 9.9796043E-04 1.00000 0.00000 sc13 8.7804442E-06 1.00000 0.00000 sc18 2.4360058E-04 1.00000 0.00000 sc19 2.0000000E-04 1.00000 0.00000 sc22 0.1989638 1.00000 0.00000 sc23 1.7940878E-05 1.00000 0.00000 Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 14: Final Calibrated Hydrographs Appendix 15: PEST Sensitivity Output COMPLETION OF OPTIMISATION PROCESS Composite sensitivities for all observations/prior info -----> Number of observations with non-zero weight = 637 Parameter Current name Group Value Sensitivity hk1 h 0.734974 4.61E-03 hk2 h 0.752222 1.84063 hk3 h 68.2525 9.71E-03 hk4 h 2.06E-03 4.14E-02 hk5 h 1.25E-02 1.95695 hk6 h 2.12E-02 3.52E-02 hk7 h 5.41E-03 1.94185 hk8 h 1.02E-03 6.59E-02 hk9 h 6.15E-02 3.48E-03 hk11 h 7.89E-02 4.51894 hk12 h 4.16E-04 1.94116 hk13 h 1.00E-05 0 hk15 h 6.23E-05 6.24E-03 hk19 h 0.11579 1.83321 hk20 h 1.00E-03 3.22E-02 hk21 h 1.04E-04 2.97E-03 vk1 v 0.738714 6.44E-03 vk2 v 1.66428 3.77E-03 vk3 v 110.984 4.80E-03 vk4 v 1.20E-02 2.18E-02 vk5 v 4.22E-03 1.69396 vk6 v 3.46E-02 2.41E-02 vk7 v 4.67E-03 1.94122 vk8 v 2.89E-03 2.28E-02 vk9 v 8.00E-02 0 vk11 v 5.20E-02 4.19E-03 vk12 v 5.18E-04 1.94261 vk13 v 1.11E-05 5.55E-03 vk14 v 2.96E-03 2.30E-02 vk15 v 4.48E-05 7.70E-03 vk16 v 0.52327 5.97E-03 vk17 v 1.13625 1.07343 vk18 v 2.00E-04 0 vk19 v 6.84E-02 1.94417 vk20 v 2.41E-04 1.94694 sy4 sy 9.71E-02 3.17E-02 sy6 sy 1.94E-02 1.06548 sy8 sy 8.66E-04 1.87304 sy22 sy 4.07586 0.90827 sc8 sc 1.13E-03 4.60E-03 sc13 sc 2.38E-05 1.25E-02 sc18 sc 3.47E-04 2.09E-02 sc19 sc 2.00E-04 0 sc22 sc 0.69288 3.54152 sc23 sc 1.25E-05 4.32E-03 Appendix 16: Prediction Model Heads Initial Heads - MTPPGM Predicted final heads - MTPPGM Appendix 17: Hydrochemistry Data Mt. Tom Price Hydrochemistry Sample List Name Easting Northing Top mRL Depth Depth mRL Type Geology Location MM01 13672.25 8186.54 712.64 90.00 622.64 S/Pipe MM MMW MM02 17642.40 10615.63 825.53 152.00 673.53 S/Pipe MM MME PZ05SSIX2A 13898.553 9183.07 660.548 83.55 577.00 Piezo MCS SSIX PZ05SSIX2B 13900.965 9182.82 660.61 97.1 563.51 Piezo MCS SSIX PZ16 15921.80 9682.40 767.79 165.00 602.79 S/Pipe WT SSIX/SEP Section 6 pond 14000 9200 660 NA NA Pit Lake DG SSIX SEP (DB3) 15847.373 10513.790 690.47 80 610.47 Bore MCS SEP WB03SEP 0001 16142.399 10539.485 615.30 60* 555.30* Bore DG SEP WB05NTD1 10393.512 12260.52 685.25 73 612.25 Bore DG NTD WB05SEP02 15904.703 10446.43 630.72 60 570.72 Bore DG SEP WB05SSIX001 13926.781 9196.36 660.06 74.8 585.26 Bore MCS SSIX WB06NTD1 10196.79 12366.88 674.9 70* 604.90* Bore DG NTD WEP03 11826.426 10892.23 750.36 80 670.36 Piezo MCS WEP WEP04 11832.82 10885.77 750.78 52 698.78 Piezo MCS WEP WEP5 12082.365 10913.14 758.01 50 708.01 Piezo MCS WEP PZ17 12995.1 8550.90 709.90 150 559.90 Piezo WT SSIX NDDW02 ? ? ? ? ? ? DG NTD PZ06WEP01 ? ? ? ? ? Piezo DG WEP PZ06MM07 ? ? ? ? ? Piezo MM MM? * estimated data Appendix 17: Hydrochemistry Data Appendix 17: Hydrochemistry Data Appendix 17: Hydrochemistry Data Appendix 17: Hydrochemistry Data Appendix 18: Time Series Acidity Trends Appendix 18: Time Series Acidity Trends Appendix 18: Time Series Acidity Trends Appendix 18: Time Series Acidity Trends Appendix 18: Time Series Acidity Trends